{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.9993145990404386,
  "eval_steps": 500,
  "global_step": 729,
  "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": 3.7571,
      "step": 1
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.333333333333334e-06,
      "loss": 3.624,
      "step": 2
    },
    {
      "epoch": 0.0,
      "learning_rate": 5.283208335737189e-06,
      "loss": 3.9204,
      "step": 3
    },
    {
      "epoch": 0.01,
      "learning_rate": 6.666666666666668e-06,
      "loss": 3.6252,
      "step": 4
    },
    {
      "epoch": 0.01,
      "learning_rate": 7.739760316291209e-06,
      "loss": 3.5256,
      "step": 5
    },
    {
      "epoch": 0.01,
      "learning_rate": 8.616541669070522e-06,
      "loss": 3.1918,
      "step": 6
    },
    {
      "epoch": 0.01,
      "learning_rate": 9.357849740192014e-06,
      "loss": 2.9474,
      "step": 7
    },
    {
      "epoch": 0.01,
      "learning_rate": 1e-05,
      "loss": 3.1106,
      "step": 8
    },
    {
      "epoch": 0.01,
      "learning_rate": 1e-05,
      "loss": 2.9399,
      "step": 9
    },
    {
      "epoch": 0.01,
      "learning_rate": 1e-05,
      "loss": 2.7671,
      "step": 10
    },
    {
      "epoch": 0.02,
      "learning_rate": 1e-05,
      "loss": 2.7104,
      "step": 11
    },
    {
      "epoch": 0.02,
      "learning_rate": 1e-05,
      "loss": 2.6174,
      "step": 12
    },
    {
      "epoch": 0.02,
      "learning_rate": 1e-05,
      "loss": 2.5677,
      "step": 13
    },
    {
      "epoch": 0.02,
      "learning_rate": 1e-05,
      "loss": 2.1427,
      "step": 14
    },
    {
      "epoch": 0.02,
      "learning_rate": 1e-05,
      "loss": 2.6958,
      "step": 15
    },
    {
      "epoch": 0.02,
      "learning_rate": 1e-05,
      "loss": 2.4929,
      "step": 16
    },
    {
      "epoch": 0.02,
      "learning_rate": 1e-05,
      "loss": 2.2651,
      "step": 17
    },
    {
      "epoch": 0.02,
      "learning_rate": 1e-05,
      "loss": 2.3931,
      "step": 18
    },
    {
      "epoch": 0.03,
      "learning_rate": 1e-05,
      "loss": 2.2779,
      "step": 19
    },
    {
      "epoch": 0.03,
      "learning_rate": 1e-05,
      "loss": 2.353,
      "step": 20
    },
    {
      "epoch": 0.03,
      "learning_rate": 1e-05,
      "loss": 2.1949,
      "step": 21
    },
    {
      "epoch": 0.03,
      "learning_rate": 1e-05,
      "loss": 2.4163,
      "step": 22
    },
    {
      "epoch": 0.03,
      "learning_rate": 1e-05,
      "loss": 2.0917,
      "step": 23
    },
    {
      "epoch": 0.03,
      "learning_rate": 1e-05,
      "loss": 2.0341,
      "step": 24
    },
    {
      "epoch": 0.03,
      "learning_rate": 1e-05,
      "loss": 1.9457,
      "step": 25
    },
    {
      "epoch": 0.04,
      "learning_rate": 1e-05,
      "loss": 2.2169,
      "step": 26
    },
    {
      "epoch": 0.04,
      "learning_rate": 1e-05,
      "loss": 2.2133,
      "step": 27
    },
    {
      "epoch": 0.04,
      "learning_rate": 1e-05,
      "loss": 2.105,
      "step": 28
    },
    {
      "epoch": 0.04,
      "learning_rate": 1e-05,
      "loss": 2.1545,
      "step": 29
    },
    {
      "epoch": 0.04,
      "learning_rate": 1e-05,
      "loss": 1.8215,
      "step": 30
    },
    {
      "epoch": 0.04,
      "learning_rate": 1e-05,
      "loss": 2.0513,
      "step": 31
    },
    {
      "epoch": 0.04,
      "learning_rate": 1e-05,
      "loss": 2.089,
      "step": 32
    },
    {
      "epoch": 0.05,
      "learning_rate": 1e-05,
      "loss": 2.1787,
      "step": 33
    },
    {
      "epoch": 0.05,
      "learning_rate": 1e-05,
      "loss": 2.0857,
      "step": 34
    },
    {
      "epoch": 0.05,
      "learning_rate": 1e-05,
      "loss": 1.8228,
      "step": 35
    },
    {
      "epoch": 0.05,
      "learning_rate": 1e-05,
      "loss": 1.7794,
      "step": 36
    },
    {
      "epoch": 0.05,
      "learning_rate": 1e-05,
      "loss": 2.1616,
      "step": 37
    },
    {
      "epoch": 0.05,
      "learning_rate": 1e-05,
      "loss": 1.9264,
      "step": 38
    },
    {
      "epoch": 0.05,
      "learning_rate": 1e-05,
      "loss": 2.04,
      "step": 39
    },
    {
      "epoch": 0.05,
      "learning_rate": 1e-05,
      "loss": 2.0891,
      "step": 40
    },
    {
      "epoch": 0.06,
      "learning_rate": 1e-05,
      "loss": 1.619,
      "step": 41
    },
    {
      "epoch": 0.06,
      "learning_rate": 1e-05,
      "loss": 2.2679,
      "step": 42
    },
    {
      "epoch": 0.06,
      "learning_rate": 1e-05,
      "loss": 1.922,
      "step": 43
    },
    {
      "epoch": 0.06,
      "learning_rate": 1e-05,
      "loss": 2.0058,
      "step": 44
    },
    {
      "epoch": 0.06,
      "learning_rate": 1e-05,
      "loss": 2.2429,
      "step": 45
    },
    {
      "epoch": 0.06,
      "learning_rate": 1e-05,
      "loss": 1.9547,
      "step": 46
    },
    {
      "epoch": 0.06,
      "learning_rate": 1e-05,
      "loss": 2.1516,
      "step": 47
    },
    {
      "epoch": 0.07,
      "learning_rate": 1e-05,
      "loss": 2.0598,
      "step": 48
    },
    {
      "epoch": 0.07,
      "learning_rate": 1e-05,
      "loss": 1.7213,
      "step": 49
    },
    {
      "epoch": 0.07,
      "learning_rate": 1e-05,
      "loss": 1.9146,
      "step": 50
    },
    {
      "epoch": 0.07,
      "learning_rate": 1e-05,
      "loss": 1.9552,
      "step": 51
    },
    {
      "epoch": 0.07,
      "learning_rate": 1e-05,
      "loss": 1.7083,
      "step": 52
    },
    {
      "epoch": 0.07,
      "learning_rate": 1e-05,
      "loss": 1.8594,
      "step": 53
    },
    {
      "epoch": 0.07,
      "learning_rate": 1e-05,
      "loss": 1.6959,
      "step": 54
    },
    {
      "epoch": 0.08,
      "learning_rate": 1e-05,
      "loss": 1.9453,
      "step": 55
    },
    {
      "epoch": 0.08,
      "learning_rate": 1e-05,
      "loss": 1.9756,
      "step": 56
    },
    {
      "epoch": 0.08,
      "learning_rate": 1e-05,
      "loss": 2.1323,
      "step": 57
    },
    {
      "epoch": 0.08,
      "learning_rate": 1e-05,
      "loss": 1.8882,
      "step": 58
    },
    {
      "epoch": 0.08,
      "learning_rate": 1e-05,
      "loss": 1.9266,
      "step": 59
    },
    {
      "epoch": 0.08,
      "learning_rate": 1e-05,
      "loss": 1.8003,
      "step": 60
    },
    {
      "epoch": 0.08,
      "learning_rate": 1e-05,
      "loss": 1.7712,
      "step": 61
    },
    {
      "epoch": 0.08,
      "learning_rate": 1e-05,
      "loss": 2.0563,
      "step": 62
    },
    {
      "epoch": 0.09,
      "learning_rate": 1e-05,
      "loss": 1.6903,
      "step": 63
    },
    {
      "epoch": 0.09,
      "learning_rate": 1e-05,
      "loss": 1.7107,
      "step": 64
    },
    {
      "epoch": 0.09,
      "learning_rate": 1e-05,
      "loss": 1.606,
      "step": 65
    },
    {
      "epoch": 0.09,
      "learning_rate": 1e-05,
      "loss": 1.7947,
      "step": 66
    },
    {
      "epoch": 0.09,
      "learning_rate": 1e-05,
      "loss": 1.8085,
      "step": 67
    },
    {
      "epoch": 0.09,
      "learning_rate": 1e-05,
      "loss": 1.8558,
      "step": 68
    },
    {
      "epoch": 0.09,
      "learning_rate": 1e-05,
      "loss": 1.8696,
      "step": 69
    },
    {
      "epoch": 0.1,
      "learning_rate": 1e-05,
      "loss": 1.6238,
      "step": 70
    },
    {
      "epoch": 0.1,
      "learning_rate": 1e-05,
      "loss": 1.7783,
      "step": 71
    },
    {
      "epoch": 0.1,
      "learning_rate": 1e-05,
      "loss": 1.6956,
      "step": 72
    },
    {
      "epoch": 0.1,
      "learning_rate": 1e-05,
      "loss": 1.8776,
      "step": 73
    },
    {
      "epoch": 0.1,
      "learning_rate": 1e-05,
      "loss": 1.8059,
      "step": 74
    },
    {
      "epoch": 0.1,
      "learning_rate": 1e-05,
      "loss": 1.721,
      "step": 75
    },
    {
      "epoch": 0.1,
      "learning_rate": 1e-05,
      "loss": 1.7748,
      "step": 76
    },
    {
      "epoch": 0.11,
      "learning_rate": 1e-05,
      "loss": 1.8159,
      "step": 77
    },
    {
      "epoch": 0.11,
      "learning_rate": 1e-05,
      "loss": 1.9375,
      "step": 78
    },
    {
      "epoch": 0.11,
      "learning_rate": 1e-05,
      "loss": 1.775,
      "step": 79
    },
    {
      "epoch": 0.11,
      "learning_rate": 1e-05,
      "loss": 1.6549,
      "step": 80
    },
    {
      "epoch": 0.11,
      "learning_rate": 1e-05,
      "loss": 1.9401,
      "step": 81
    },
    {
      "epoch": 0.11,
      "learning_rate": 1e-05,
      "loss": 1.8199,
      "step": 82
    },
    {
      "epoch": 0.11,
      "learning_rate": 1e-05,
      "loss": 1.6188,
      "step": 83
    },
    {
      "epoch": 0.12,
      "learning_rate": 1e-05,
      "loss": 1.7208,
      "step": 84
    },
    {
      "epoch": 0.12,
      "learning_rate": 1e-05,
      "loss": 1.881,
      "step": 85
    },
    {
      "epoch": 0.12,
      "learning_rate": 1e-05,
      "loss": 1.5557,
      "step": 86
    },
    {
      "epoch": 0.12,
      "learning_rate": 1e-05,
      "loss": 1.7011,
      "step": 87
    },
    {
      "epoch": 0.12,
      "learning_rate": 1e-05,
      "loss": 1.9393,
      "step": 88
    },
    {
      "epoch": 0.12,
      "learning_rate": 1e-05,
      "loss": 1.8495,
      "step": 89
    },
    {
      "epoch": 0.12,
      "learning_rate": 1e-05,
      "loss": 1.9412,
      "step": 90
    },
    {
      "epoch": 0.12,
      "learning_rate": 1e-05,
      "loss": 1.6088,
      "step": 91
    },
    {
      "epoch": 0.13,
      "learning_rate": 1e-05,
      "loss": 1.7219,
      "step": 92
    },
    {
      "epoch": 0.13,
      "learning_rate": 1e-05,
      "loss": 1.9291,
      "step": 93
    },
    {
      "epoch": 0.13,
      "learning_rate": 1e-05,
      "loss": 2.1837,
      "step": 94
    },
    {
      "epoch": 0.13,
      "learning_rate": 1e-05,
      "loss": 1.6232,
      "step": 95
    },
    {
      "epoch": 0.13,
      "learning_rate": 1e-05,
      "loss": 1.6476,
      "step": 96
    },
    {
      "epoch": 0.13,
      "learning_rate": 1e-05,
      "loss": 1.6934,
      "step": 97
    },
    {
      "epoch": 0.13,
      "learning_rate": 1e-05,
      "loss": 1.7029,
      "step": 98
    },
    {
      "epoch": 0.14,
      "learning_rate": 1e-05,
      "loss": 1.8186,
      "step": 99
    },
    {
      "epoch": 0.14,
      "learning_rate": 1e-05,
      "loss": 1.8478,
      "step": 100
    },
    {
      "epoch": 0.14,
      "learning_rate": 1e-05,
      "loss": 1.4334,
      "step": 101
    },
    {
      "epoch": 0.14,
      "learning_rate": 1e-05,
      "loss": 1.7916,
      "step": 102
    },
    {
      "epoch": 0.14,
      "learning_rate": 1e-05,
      "loss": 1.8307,
      "step": 103
    },
    {
      "epoch": 0.14,
      "learning_rate": 1e-05,
      "loss": 1.7422,
      "step": 104
    },
    {
      "epoch": 0.14,
      "learning_rate": 1e-05,
      "loss": 1.615,
      "step": 105
    },
    {
      "epoch": 0.15,
      "learning_rate": 1e-05,
      "loss": 1.6585,
      "step": 106
    },
    {
      "epoch": 0.15,
      "learning_rate": 1e-05,
      "loss": 1.9481,
      "step": 107
    },
    {
      "epoch": 0.15,
      "learning_rate": 1e-05,
      "loss": 1.5223,
      "step": 108
    },
    {
      "epoch": 0.15,
      "learning_rate": 1e-05,
      "loss": 1.8279,
      "step": 109
    },
    {
      "epoch": 0.15,
      "learning_rate": 1e-05,
      "loss": 1.5541,
      "step": 110
    },
    {
      "epoch": 0.15,
      "learning_rate": 1e-05,
      "loss": 1.8175,
      "step": 111
    },
    {
      "epoch": 0.15,
      "learning_rate": 1e-05,
      "loss": 1.5625,
      "step": 112
    },
    {
      "epoch": 0.15,
      "learning_rate": 1e-05,
      "loss": 1.983,
      "step": 113
    },
    {
      "epoch": 0.16,
      "learning_rate": 1e-05,
      "loss": 1.64,
      "step": 114
    },
    {
      "epoch": 0.16,
      "learning_rate": 1e-05,
      "loss": 1.5634,
      "step": 115
    },
    {
      "epoch": 0.16,
      "learning_rate": 1e-05,
      "loss": 1.811,
      "step": 116
    },
    {
      "epoch": 0.16,
      "learning_rate": 1e-05,
      "loss": 1.9193,
      "step": 117
    },
    {
      "epoch": 0.16,
      "learning_rate": 1e-05,
      "loss": 1.3957,
      "step": 118
    },
    {
      "epoch": 0.16,
      "learning_rate": 1e-05,
      "loss": 1.4065,
      "step": 119
    },
    {
      "epoch": 0.16,
      "learning_rate": 1e-05,
      "loss": 1.8184,
      "step": 120
    },
    {
      "epoch": 0.17,
      "learning_rate": 1e-05,
      "loss": 1.6816,
      "step": 121
    },
    {
      "epoch": 0.17,
      "learning_rate": 1e-05,
      "loss": 1.7883,
      "step": 122
    },
    {
      "epoch": 0.17,
      "learning_rate": 1e-05,
      "loss": 1.2636,
      "step": 123
    },
    {
      "epoch": 0.17,
      "learning_rate": 1e-05,
      "loss": 1.7257,
      "step": 124
    },
    {
      "epoch": 0.17,
      "learning_rate": 1e-05,
      "loss": 1.8263,
      "step": 125
    },
    {
      "epoch": 0.17,
      "learning_rate": 1e-05,
      "loss": 1.7815,
      "step": 126
    },
    {
      "epoch": 0.17,
      "learning_rate": 1e-05,
      "loss": 1.4648,
      "step": 127
    },
    {
      "epoch": 0.18,
      "learning_rate": 1e-05,
      "loss": 1.7935,
      "step": 128
    },
    {
      "epoch": 0.18,
      "learning_rate": 1e-05,
      "loss": 1.7328,
      "step": 129
    },
    {
      "epoch": 0.18,
      "learning_rate": 1e-05,
      "loss": 1.6547,
      "step": 130
    },
    {
      "epoch": 0.18,
      "learning_rate": 1e-05,
      "loss": 1.5887,
      "step": 131
    },
    {
      "epoch": 0.18,
      "learning_rate": 1e-05,
      "loss": 1.7388,
      "step": 132
    },
    {
      "epoch": 0.18,
      "learning_rate": 1e-05,
      "loss": 1.6904,
      "step": 133
    },
    {
      "epoch": 0.18,
      "learning_rate": 1e-05,
      "loss": 2.0132,
      "step": 134
    },
    {
      "epoch": 0.19,
      "learning_rate": 1e-05,
      "loss": 1.5199,
      "step": 135
    },
    {
      "epoch": 0.19,
      "learning_rate": 1e-05,
      "loss": 1.476,
      "step": 136
    },
    {
      "epoch": 0.19,
      "learning_rate": 1e-05,
      "loss": 1.5699,
      "step": 137
    },
    {
      "epoch": 0.19,
      "learning_rate": 1e-05,
      "loss": 1.7587,
      "step": 138
    },
    {
      "epoch": 0.19,
      "learning_rate": 1e-05,
      "loss": 1.6595,
      "step": 139
    },
    {
      "epoch": 0.19,
      "learning_rate": 1e-05,
      "loss": 1.5561,
      "step": 140
    },
    {
      "epoch": 0.19,
      "learning_rate": 1e-05,
      "loss": 1.6922,
      "step": 141
    },
    {
      "epoch": 0.19,
      "learning_rate": 1e-05,
      "loss": 1.619,
      "step": 142
    },
    {
      "epoch": 0.2,
      "learning_rate": 1e-05,
      "loss": 1.6969,
      "step": 143
    },
    {
      "epoch": 0.2,
      "learning_rate": 1e-05,
      "loss": 1.7635,
      "step": 144
    },
    {
      "epoch": 0.2,
      "learning_rate": 1e-05,
      "loss": 1.4982,
      "step": 145
    },
    {
      "epoch": 0.2,
      "learning_rate": 1e-05,
      "loss": 1.5431,
      "step": 146
    },
    {
      "epoch": 0.2,
      "learning_rate": 1e-05,
      "loss": 1.5194,
      "step": 147
    },
    {
      "epoch": 0.2,
      "learning_rate": 1e-05,
      "loss": 1.5283,
      "step": 148
    },
    {
      "epoch": 0.2,
      "learning_rate": 1e-05,
      "loss": 1.4764,
      "step": 149
    },
    {
      "epoch": 0.21,
      "learning_rate": 1e-05,
      "loss": 1.7428,
      "step": 150
    },
    {
      "epoch": 0.21,
      "learning_rate": 1e-05,
      "loss": 1.5739,
      "step": 151
    },
    {
      "epoch": 0.21,
      "learning_rate": 1e-05,
      "loss": 1.9455,
      "step": 152
    },
    {
      "epoch": 0.21,
      "learning_rate": 1e-05,
      "loss": 1.4836,
      "step": 153
    },
    {
      "epoch": 0.21,
      "learning_rate": 1e-05,
      "loss": 1.4432,
      "step": 154
    },
    {
      "epoch": 0.21,
      "learning_rate": 1e-05,
      "loss": 1.6208,
      "step": 155
    },
    {
      "epoch": 0.21,
      "learning_rate": 1e-05,
      "loss": 1.3398,
      "step": 156
    },
    {
      "epoch": 0.22,
      "learning_rate": 1e-05,
      "loss": 1.6181,
      "step": 157
    },
    {
      "epoch": 0.22,
      "learning_rate": 1e-05,
      "loss": 1.4816,
      "step": 158
    },
    {
      "epoch": 0.22,
      "learning_rate": 1e-05,
      "loss": 1.7345,
      "step": 159
    },
    {
      "epoch": 0.22,
      "learning_rate": 1e-05,
      "loss": 1.6008,
      "step": 160
    },
    {
      "epoch": 0.22,
      "learning_rate": 1e-05,
      "loss": 1.4235,
      "step": 161
    },
    {
      "epoch": 0.22,
      "learning_rate": 1e-05,
      "loss": 1.4032,
      "step": 162
    },
    {
      "epoch": 0.22,
      "learning_rate": 1e-05,
      "loss": 1.6597,
      "step": 163
    },
    {
      "epoch": 0.22,
      "learning_rate": 1e-05,
      "loss": 1.5719,
      "step": 164
    },
    {
      "epoch": 0.23,
      "learning_rate": 1e-05,
      "loss": 1.6354,
      "step": 165
    },
    {
      "epoch": 0.23,
      "learning_rate": 1e-05,
      "loss": 1.4573,
      "step": 166
    },
    {
      "epoch": 0.23,
      "learning_rate": 1e-05,
      "loss": 1.5803,
      "step": 167
    },
    {
      "epoch": 0.23,
      "learning_rate": 1e-05,
      "loss": 1.658,
      "step": 168
    },
    {
      "epoch": 0.23,
      "learning_rate": 1e-05,
      "loss": 1.5798,
      "step": 169
    },
    {
      "epoch": 0.23,
      "learning_rate": 1e-05,
      "loss": 1.4604,
      "step": 170
    },
    {
      "epoch": 0.23,
      "learning_rate": 1e-05,
      "loss": 1.5803,
      "step": 171
    },
    {
      "epoch": 0.24,
      "learning_rate": 1e-05,
      "loss": 1.611,
      "step": 172
    },
    {
      "epoch": 0.24,
      "learning_rate": 1e-05,
      "loss": 1.2751,
      "step": 173
    },
    {
      "epoch": 0.24,
      "learning_rate": 1e-05,
      "loss": 1.5076,
      "step": 174
    },
    {
      "epoch": 0.24,
      "learning_rate": 1e-05,
      "loss": 1.5532,
      "step": 175
    },
    {
      "epoch": 0.24,
      "learning_rate": 1e-05,
      "loss": 1.6382,
      "step": 176
    },
    {
      "epoch": 0.24,
      "learning_rate": 1e-05,
      "loss": 1.5485,
      "step": 177
    },
    {
      "epoch": 0.24,
      "learning_rate": 1e-05,
      "loss": 1.395,
      "step": 178
    },
    {
      "epoch": 0.25,
      "learning_rate": 1e-05,
      "loss": 1.5505,
      "step": 179
    },
    {
      "epoch": 0.25,
      "learning_rate": 1e-05,
      "loss": 1.6763,
      "step": 180
    },
    {
      "epoch": 0.25,
      "learning_rate": 1e-05,
      "loss": 1.6832,
      "step": 181
    },
    {
      "epoch": 0.25,
      "learning_rate": 1e-05,
      "loss": 1.7668,
      "step": 182
    },
    {
      "epoch": 0.25,
      "learning_rate": 1e-05,
      "loss": 1.659,
      "step": 183
    },
    {
      "epoch": 0.25,
      "learning_rate": 1e-05,
      "loss": 1.8761,
      "step": 184
    },
    {
      "epoch": 0.25,
      "learning_rate": 1e-05,
      "loss": 1.7204,
      "step": 185
    },
    {
      "epoch": 0.25,
      "learning_rate": 1e-05,
      "loss": 1.7615,
      "step": 186
    },
    {
      "epoch": 0.26,
      "learning_rate": 1e-05,
      "loss": 1.6719,
      "step": 187
    },
    {
      "epoch": 0.26,
      "learning_rate": 1e-05,
      "loss": 1.6348,
      "step": 188
    },
    {
      "epoch": 0.26,
      "learning_rate": 1e-05,
      "loss": 1.8078,
      "step": 189
    },
    {
      "epoch": 0.26,
      "learning_rate": 1e-05,
      "loss": 1.5327,
      "step": 190
    },
    {
      "epoch": 0.26,
      "learning_rate": 1e-05,
      "loss": 1.736,
      "step": 191
    },
    {
      "epoch": 0.26,
      "learning_rate": 1e-05,
      "loss": 1.441,
      "step": 192
    },
    {
      "epoch": 0.26,
      "learning_rate": 1e-05,
      "loss": 1.5573,
      "step": 193
    },
    {
      "epoch": 0.27,
      "learning_rate": 1e-05,
      "loss": 1.4477,
      "step": 194
    },
    {
      "epoch": 0.27,
      "learning_rate": 1e-05,
      "loss": 1.502,
      "step": 195
    },
    {
      "epoch": 0.27,
      "learning_rate": 1e-05,
      "loss": 1.6167,
      "step": 196
    },
    {
      "epoch": 0.27,
      "learning_rate": 1e-05,
      "loss": 1.54,
      "step": 197
    },
    {
      "epoch": 0.27,
      "learning_rate": 1e-05,
      "loss": 1.6512,
      "step": 198
    },
    {
      "epoch": 0.27,
      "learning_rate": 1e-05,
      "loss": 1.4501,
      "step": 199
    },
    {
      "epoch": 0.27,
      "learning_rate": 1e-05,
      "loss": 1.2765,
      "step": 200
    },
    {
      "epoch": 0.28,
      "learning_rate": 1e-05,
      "loss": 1.408,
      "step": 201
    },
    {
      "epoch": 0.28,
      "learning_rate": 1e-05,
      "loss": 1.5561,
      "step": 202
    },
    {
      "epoch": 0.28,
      "learning_rate": 1e-05,
      "loss": 1.643,
      "step": 203
    },
    {
      "epoch": 0.28,
      "learning_rate": 1e-05,
      "loss": 1.379,
      "step": 204
    },
    {
      "epoch": 0.28,
      "learning_rate": 1e-05,
      "loss": 1.5696,
      "step": 205
    },
    {
      "epoch": 0.28,
      "learning_rate": 1e-05,
      "loss": 1.4753,
      "step": 206
    },
    {
      "epoch": 0.28,
      "learning_rate": 1e-05,
      "loss": 1.7334,
      "step": 207
    },
    {
      "epoch": 0.29,
      "learning_rate": 1e-05,
      "loss": 1.393,
      "step": 208
    },
    {
      "epoch": 0.29,
      "learning_rate": 1e-05,
      "loss": 1.5177,
      "step": 209
    },
    {
      "epoch": 0.29,
      "learning_rate": 1e-05,
      "loss": 1.5547,
      "step": 210
    },
    {
      "epoch": 0.29,
      "learning_rate": 1e-05,
      "loss": 1.597,
      "step": 211
    },
    {
      "epoch": 0.29,
      "learning_rate": 1e-05,
      "loss": 1.5884,
      "step": 212
    },
    {
      "epoch": 0.29,
      "learning_rate": 1e-05,
      "loss": 1.5523,
      "step": 213
    },
    {
      "epoch": 0.29,
      "learning_rate": 1e-05,
      "loss": 1.4321,
      "step": 214
    },
    {
      "epoch": 0.29,
      "learning_rate": 1e-05,
      "loss": 1.5629,
      "step": 215
    },
    {
      "epoch": 0.3,
      "learning_rate": 1e-05,
      "loss": 1.545,
      "step": 216
    },
    {
      "epoch": 0.3,
      "learning_rate": 1e-05,
      "loss": 1.4742,
      "step": 217
    },
    {
      "epoch": 0.3,
      "learning_rate": 1e-05,
      "loss": 1.4188,
      "step": 218
    },
    {
      "epoch": 0.3,
      "learning_rate": 1e-05,
      "loss": 1.4225,
      "step": 219
    },
    {
      "epoch": 0.3,
      "learning_rate": 1e-05,
      "loss": 1.5398,
      "step": 220
    },
    {
      "epoch": 0.3,
      "learning_rate": 1e-05,
      "loss": 1.6857,
      "step": 221
    },
    {
      "epoch": 0.3,
      "learning_rate": 1e-05,
      "loss": 1.6039,
      "step": 222
    },
    {
      "epoch": 0.31,
      "learning_rate": 1e-05,
      "loss": 1.375,
      "step": 223
    },
    {
      "epoch": 0.31,
      "learning_rate": 1e-05,
      "loss": 1.5747,
      "step": 224
    },
    {
      "epoch": 0.31,
      "learning_rate": 1e-05,
      "loss": 1.6246,
      "step": 225
    },
    {
      "epoch": 0.31,
      "learning_rate": 1e-05,
      "loss": 1.4102,
      "step": 226
    },
    {
      "epoch": 0.31,
      "learning_rate": 1e-05,
      "loss": 1.4295,
      "step": 227
    },
    {
      "epoch": 0.31,
      "learning_rate": 1e-05,
      "loss": 1.3068,
      "step": 228
    },
    {
      "epoch": 0.31,
      "learning_rate": 1e-05,
      "loss": 1.5571,
      "step": 229
    },
    {
      "epoch": 0.32,
      "learning_rate": 1e-05,
      "loss": 1.5746,
      "step": 230
    },
    {
      "epoch": 0.32,
      "learning_rate": 1e-05,
      "loss": 1.6786,
      "step": 231
    },
    {
      "epoch": 0.32,
      "learning_rate": 1e-05,
      "loss": 1.6725,
      "step": 232
    },
    {
      "epoch": 0.32,
      "learning_rate": 1e-05,
      "loss": 1.3939,
      "step": 233
    },
    {
      "epoch": 0.32,
      "learning_rate": 1e-05,
      "loss": 1.1849,
      "step": 234
    },
    {
      "epoch": 0.32,
      "learning_rate": 1e-05,
      "loss": 1.46,
      "step": 235
    },
    {
      "epoch": 0.32,
      "learning_rate": 1e-05,
      "loss": 1.3777,
      "step": 236
    },
    {
      "epoch": 0.32,
      "learning_rate": 1e-05,
      "loss": 1.7902,
      "step": 237
    },
    {
      "epoch": 0.33,
      "learning_rate": 1e-05,
      "loss": 1.4057,
      "step": 238
    },
    {
      "epoch": 0.33,
      "learning_rate": 1e-05,
      "loss": 1.7069,
      "step": 239
    },
    {
      "epoch": 0.33,
      "learning_rate": 1e-05,
      "loss": 1.6913,
      "step": 240
    },
    {
      "epoch": 0.33,
      "learning_rate": 1e-05,
      "loss": 1.7906,
      "step": 241
    },
    {
      "epoch": 0.33,
      "learning_rate": 1e-05,
      "loss": 1.1876,
      "step": 242
    },
    {
      "epoch": 0.33,
      "learning_rate": 1e-05,
      "loss": 1.6128,
      "step": 243
    },
    {
      "epoch": 0.33,
      "learning_rate": 1e-05,
      "loss": 1.5079,
      "step": 244
    },
    {
      "epoch": 0.34,
      "learning_rate": 1e-05,
      "loss": 1.4811,
      "step": 245
    },
    {
      "epoch": 0.34,
      "learning_rate": 1e-05,
      "loss": 1.2664,
      "step": 246
    },
    {
      "epoch": 0.34,
      "learning_rate": 1e-05,
      "loss": 1.5449,
      "step": 247
    },
    {
      "epoch": 0.34,
      "learning_rate": 1e-05,
      "loss": 1.264,
      "step": 248
    },
    {
      "epoch": 0.34,
      "learning_rate": 1e-05,
      "loss": 1.4283,
      "step": 249
    },
    {
      "epoch": 0.34,
      "learning_rate": 1e-05,
      "loss": 1.4509,
      "step": 250
    },
    {
      "epoch": 0.34,
      "learning_rate": 1e-05,
      "loss": 1.6738,
      "step": 251
    },
    {
      "epoch": 0.35,
      "learning_rate": 1e-05,
      "loss": 1.4787,
      "step": 252
    },
    {
      "epoch": 0.35,
      "learning_rate": 1e-05,
      "loss": 1.5503,
      "step": 253
    },
    {
      "epoch": 0.35,
      "learning_rate": 1e-05,
      "loss": 1.6582,
      "step": 254
    },
    {
      "epoch": 0.35,
      "learning_rate": 1e-05,
      "loss": 1.4816,
      "step": 255
    },
    {
      "epoch": 0.35,
      "learning_rate": 1e-05,
      "loss": 1.2487,
      "step": 256
    },
    {
      "epoch": 0.35,
      "learning_rate": 1e-05,
      "loss": 1.615,
      "step": 257
    },
    {
      "epoch": 0.35,
      "learning_rate": 1e-05,
      "loss": 1.3042,
      "step": 258
    },
    {
      "epoch": 0.36,
      "learning_rate": 1e-05,
      "loss": 1.7656,
      "step": 259
    },
    {
      "epoch": 0.36,
      "learning_rate": 1e-05,
      "loss": 1.2684,
      "step": 260
    },
    {
      "epoch": 0.36,
      "learning_rate": 1e-05,
      "loss": 1.368,
      "step": 261
    },
    {
      "epoch": 0.36,
      "learning_rate": 1e-05,
      "loss": 1.3118,
      "step": 262
    },
    {
      "epoch": 0.36,
      "learning_rate": 1e-05,
      "loss": 1.3155,
      "step": 263
    },
    {
      "epoch": 0.36,
      "learning_rate": 1e-05,
      "loss": 1.3507,
      "step": 264
    },
    {
      "epoch": 0.36,
      "learning_rate": 1e-05,
      "loss": 1.3469,
      "step": 265
    },
    {
      "epoch": 0.36,
      "learning_rate": 1e-05,
      "loss": 1.6111,
      "step": 266
    },
    {
      "epoch": 0.37,
      "learning_rate": 1e-05,
      "loss": 1.4193,
      "step": 267
    },
    {
      "epoch": 0.37,
      "learning_rate": 1e-05,
      "loss": 1.166,
      "step": 268
    },
    {
      "epoch": 0.37,
      "learning_rate": 1e-05,
      "loss": 1.1452,
      "step": 269
    },
    {
      "epoch": 0.37,
      "learning_rate": 1e-05,
      "loss": 1.463,
      "step": 270
    },
    {
      "epoch": 0.37,
      "learning_rate": 1e-05,
      "loss": 1.4366,
      "step": 271
    },
    {
      "epoch": 0.37,
      "learning_rate": 1e-05,
      "loss": 1.3703,
      "step": 272
    },
    {
      "epoch": 0.37,
      "learning_rate": 1e-05,
      "loss": 1.3085,
      "step": 273
    },
    {
      "epoch": 0.38,
      "learning_rate": 1e-05,
      "loss": 1.4733,
      "step": 274
    },
    {
      "epoch": 0.38,
      "learning_rate": 1e-05,
      "loss": 1.2689,
      "step": 275
    },
    {
      "epoch": 0.38,
      "learning_rate": 1e-05,
      "loss": 1.0594,
      "step": 276
    },
    {
      "epoch": 0.38,
      "learning_rate": 1e-05,
      "loss": 1.3486,
      "step": 277
    },
    {
      "epoch": 0.38,
      "learning_rate": 1e-05,
      "loss": 1.3228,
      "step": 278
    },
    {
      "epoch": 0.38,
      "learning_rate": 1e-05,
      "loss": 1.3364,
      "step": 279
    },
    {
      "epoch": 0.38,
      "learning_rate": 1e-05,
      "loss": 1.4263,
      "step": 280
    },
    {
      "epoch": 0.39,
      "learning_rate": 1e-05,
      "loss": 1.2621,
      "step": 281
    },
    {
      "epoch": 0.39,
      "learning_rate": 1e-05,
      "loss": 1.3835,
      "step": 282
    },
    {
      "epoch": 0.39,
      "learning_rate": 1e-05,
      "loss": 1.4006,
      "step": 283
    },
    {
      "epoch": 0.39,
      "learning_rate": 1e-05,
      "loss": 1.6985,
      "step": 284
    },
    {
      "epoch": 0.39,
      "learning_rate": 1e-05,
      "loss": 1.1841,
      "step": 285
    },
    {
      "epoch": 0.39,
      "learning_rate": 1e-05,
      "loss": 1.3197,
      "step": 286
    },
    {
      "epoch": 0.39,
      "learning_rate": 1e-05,
      "loss": 1.6785,
      "step": 287
    },
    {
      "epoch": 0.39,
      "learning_rate": 1e-05,
      "loss": 1.446,
      "step": 288
    },
    {
      "epoch": 0.4,
      "learning_rate": 1e-05,
      "loss": 1.4342,
      "step": 289
    },
    {
      "epoch": 0.4,
      "learning_rate": 1e-05,
      "loss": 1.535,
      "step": 290
    },
    {
      "epoch": 0.4,
      "learning_rate": 1e-05,
      "loss": 1.4971,
      "step": 291
    },
    {
      "epoch": 0.4,
      "learning_rate": 1e-05,
      "loss": 1.4941,
      "step": 292
    },
    {
      "epoch": 0.4,
      "learning_rate": 1e-05,
      "loss": 1.3675,
      "step": 293
    },
    {
      "epoch": 0.4,
      "learning_rate": 1e-05,
      "loss": 1.5556,
      "step": 294
    },
    {
      "epoch": 0.4,
      "learning_rate": 1e-05,
      "loss": 1.4217,
      "step": 295
    },
    {
      "epoch": 0.41,
      "learning_rate": 1e-05,
      "loss": 1.3199,
      "step": 296
    },
    {
      "epoch": 0.41,
      "learning_rate": 1e-05,
      "loss": 1.3854,
      "step": 297
    },
    {
      "epoch": 0.41,
      "learning_rate": 1e-05,
      "loss": 1.4666,
      "step": 298
    },
    {
      "epoch": 0.41,
      "learning_rate": 1e-05,
      "loss": 1.4939,
      "step": 299
    },
    {
      "epoch": 0.41,
      "learning_rate": 1e-05,
      "loss": 1.3245,
      "step": 300
    },
    {
      "epoch": 0.41,
      "learning_rate": 1e-05,
      "loss": 1.3592,
      "step": 301
    },
    {
      "epoch": 0.41,
      "learning_rate": 1e-05,
      "loss": 1.4313,
      "step": 302
    },
    {
      "epoch": 0.42,
      "learning_rate": 1e-05,
      "loss": 1.5373,
      "step": 303
    },
    {
      "epoch": 0.42,
      "learning_rate": 1e-05,
      "loss": 1.4095,
      "step": 304
    },
    {
      "epoch": 0.42,
      "learning_rate": 1e-05,
      "loss": 1.125,
      "step": 305
    },
    {
      "epoch": 0.42,
      "learning_rate": 1e-05,
      "loss": 1.4698,
      "step": 306
    },
    {
      "epoch": 0.42,
      "learning_rate": 1e-05,
      "loss": 1.366,
      "step": 307
    },
    {
      "epoch": 0.42,
      "learning_rate": 1e-05,
      "loss": 1.1242,
      "step": 308
    },
    {
      "epoch": 0.42,
      "learning_rate": 1e-05,
      "loss": 1.5119,
      "step": 309
    },
    {
      "epoch": 0.42,
      "learning_rate": 1e-05,
      "loss": 1.4328,
      "step": 310
    },
    {
      "epoch": 0.43,
      "learning_rate": 1e-05,
      "loss": 1.5072,
      "step": 311
    },
    {
      "epoch": 0.43,
      "learning_rate": 1e-05,
      "loss": 1.4553,
      "step": 312
    },
    {
      "epoch": 0.43,
      "learning_rate": 1e-05,
      "loss": 1.3752,
      "step": 313
    },
    {
      "epoch": 0.43,
      "learning_rate": 1e-05,
      "loss": 1.468,
      "step": 314
    },
    {
      "epoch": 0.43,
      "learning_rate": 1e-05,
      "loss": 1.3314,
      "step": 315
    },
    {
      "epoch": 0.43,
      "learning_rate": 1e-05,
      "loss": 1.2516,
      "step": 316
    },
    {
      "epoch": 0.43,
      "learning_rate": 1e-05,
      "loss": 1.1767,
      "step": 317
    },
    {
      "epoch": 0.44,
      "learning_rate": 1e-05,
      "loss": 1.4673,
      "step": 318
    },
    {
      "epoch": 0.44,
      "learning_rate": 1e-05,
      "loss": 1.2966,
      "step": 319
    },
    {
      "epoch": 0.44,
      "learning_rate": 1e-05,
      "loss": 1.5219,
      "step": 320
    },
    {
      "epoch": 0.44,
      "learning_rate": 1e-05,
      "loss": 1.3888,
      "step": 321
    },
    {
      "epoch": 0.44,
      "learning_rate": 1e-05,
      "loss": 1.1928,
      "step": 322
    },
    {
      "epoch": 0.44,
      "learning_rate": 1e-05,
      "loss": 1.2237,
      "step": 323
    },
    {
      "epoch": 0.44,
      "learning_rate": 1e-05,
      "loss": 1.5836,
      "step": 324
    },
    {
      "epoch": 0.45,
      "learning_rate": 1e-05,
      "loss": 1.3396,
      "step": 325
    },
    {
      "epoch": 0.45,
      "learning_rate": 1e-05,
      "loss": 1.3447,
      "step": 326
    },
    {
      "epoch": 0.45,
      "learning_rate": 1e-05,
      "loss": 1.7417,
      "step": 327
    },
    {
      "epoch": 0.45,
      "learning_rate": 1e-05,
      "loss": 1.3853,
      "step": 328
    },
    {
      "epoch": 0.45,
      "learning_rate": 1e-05,
      "loss": 1.3199,
      "step": 329
    },
    {
      "epoch": 0.45,
      "learning_rate": 1e-05,
      "loss": 1.4855,
      "step": 330
    },
    {
      "epoch": 0.45,
      "learning_rate": 1e-05,
      "loss": 1.3145,
      "step": 331
    },
    {
      "epoch": 0.46,
      "learning_rate": 1e-05,
      "loss": 1.3693,
      "step": 332
    },
    {
      "epoch": 0.46,
      "learning_rate": 1e-05,
      "loss": 1.2076,
      "step": 333
    },
    {
      "epoch": 0.46,
      "learning_rate": 1e-05,
      "loss": 1.2921,
      "step": 334
    },
    {
      "epoch": 0.46,
      "learning_rate": 1e-05,
      "loss": 1.0672,
      "step": 335
    },
    {
      "epoch": 0.46,
      "learning_rate": 1e-05,
      "loss": 1.3637,
      "step": 336
    },
    {
      "epoch": 0.46,
      "learning_rate": 1e-05,
      "loss": 1.4431,
      "step": 337
    },
    {
      "epoch": 0.46,
      "learning_rate": 1e-05,
      "loss": 1.5392,
      "step": 338
    },
    {
      "epoch": 0.46,
      "learning_rate": 1e-05,
      "loss": 1.0652,
      "step": 339
    },
    {
      "epoch": 0.47,
      "learning_rate": 1e-05,
      "loss": 1.3007,
      "step": 340
    },
    {
      "epoch": 0.47,
      "learning_rate": 1e-05,
      "loss": 1.2717,
      "step": 341
    },
    {
      "epoch": 0.47,
      "learning_rate": 1e-05,
      "loss": 1.5882,
      "step": 342
    },
    {
      "epoch": 0.47,
      "learning_rate": 1e-05,
      "loss": 1.4684,
      "step": 343
    },
    {
      "epoch": 0.47,
      "learning_rate": 1e-05,
      "loss": 1.2436,
      "step": 344
    },
    {
      "epoch": 0.47,
      "learning_rate": 1e-05,
      "loss": 1.4324,
      "step": 345
    },
    {
      "epoch": 0.47,
      "learning_rate": 1e-05,
      "loss": 1.2346,
      "step": 346
    },
    {
      "epoch": 0.48,
      "learning_rate": 1e-05,
      "loss": 1.3291,
      "step": 347
    },
    {
      "epoch": 0.48,
      "learning_rate": 1e-05,
      "loss": 1.3582,
      "step": 348
    },
    {
      "epoch": 0.48,
      "learning_rate": 1e-05,
      "loss": 1.3631,
      "step": 349
    },
    {
      "epoch": 0.48,
      "learning_rate": 1e-05,
      "loss": 1.3616,
      "step": 350
    },
    {
      "epoch": 0.48,
      "learning_rate": 1e-05,
      "loss": 1.3047,
      "step": 351
    },
    {
      "epoch": 0.48,
      "learning_rate": 1e-05,
      "loss": 1.4813,
      "step": 352
    },
    {
      "epoch": 0.48,
      "learning_rate": 1e-05,
      "loss": 1.6418,
      "step": 353
    },
    {
      "epoch": 0.49,
      "learning_rate": 1e-05,
      "loss": 1.3388,
      "step": 354
    },
    {
      "epoch": 0.49,
      "learning_rate": 1e-05,
      "loss": 1.0211,
      "step": 355
    },
    {
      "epoch": 0.49,
      "learning_rate": 1e-05,
      "loss": 1.5099,
      "step": 356
    },
    {
      "epoch": 0.49,
      "learning_rate": 1e-05,
      "loss": 1.3719,
      "step": 357
    },
    {
      "epoch": 0.49,
      "learning_rate": 1e-05,
      "loss": 1.3137,
      "step": 358
    },
    {
      "epoch": 0.49,
      "learning_rate": 1e-05,
      "loss": 1.4402,
      "step": 359
    },
    {
      "epoch": 0.49,
      "learning_rate": 1e-05,
      "loss": 1.4835,
      "step": 360
    },
    {
      "epoch": 0.49,
      "learning_rate": 1e-05,
      "loss": 1.2177,
      "step": 361
    },
    {
      "epoch": 0.5,
      "learning_rate": 1e-05,
      "loss": 1.3922,
      "step": 362
    },
    {
      "epoch": 0.5,
      "learning_rate": 1e-05,
      "loss": 1.3434,
      "step": 363
    },
    {
      "epoch": 0.5,
      "learning_rate": 1e-05,
      "loss": 1.4164,
      "step": 364
    },
    {
      "epoch": 0.5,
      "learning_rate": 1e-05,
      "loss": 1.1188,
      "step": 365
    },
    {
      "epoch": 0.5,
      "learning_rate": 1e-05,
      "loss": 1.1328,
      "step": 366
    },
    {
      "epoch": 0.5,
      "learning_rate": 1e-05,
      "loss": 1.2523,
      "step": 367
    },
    {
      "epoch": 0.5,
      "learning_rate": 1e-05,
      "loss": 1.545,
      "step": 368
    },
    {
      "epoch": 0.51,
      "learning_rate": 1e-05,
      "loss": 1.4388,
      "step": 369
    },
    {
      "epoch": 0.51,
      "learning_rate": 1e-05,
      "loss": 1.4793,
      "step": 370
    },
    {
      "epoch": 0.51,
      "learning_rate": 1e-05,
      "loss": 1.1976,
      "step": 371
    },
    {
      "epoch": 0.51,
      "learning_rate": 1e-05,
      "loss": 1.3541,
      "step": 372
    },
    {
      "epoch": 0.51,
      "learning_rate": 1e-05,
      "loss": 1.4752,
      "step": 373
    },
    {
      "epoch": 0.51,
      "learning_rate": 1e-05,
      "loss": 1.4665,
      "step": 374
    },
    {
      "epoch": 0.51,
      "learning_rate": 1e-05,
      "loss": 1.3125,
      "step": 375
    },
    {
      "epoch": 0.52,
      "learning_rate": 1e-05,
      "loss": 1.2853,
      "step": 376
    },
    {
      "epoch": 0.52,
      "learning_rate": 1e-05,
      "loss": 1.5306,
      "step": 377
    },
    {
      "epoch": 0.52,
      "learning_rate": 1e-05,
      "loss": 1.4491,
      "step": 378
    },
    {
      "epoch": 0.52,
      "learning_rate": 1e-05,
      "loss": 1.3014,
      "step": 379
    },
    {
      "epoch": 0.52,
      "learning_rate": 1e-05,
      "loss": 1.5201,
      "step": 380
    },
    {
      "epoch": 0.52,
      "learning_rate": 1e-05,
      "loss": 1.4558,
      "step": 381
    },
    {
      "epoch": 0.52,
      "learning_rate": 1e-05,
      "loss": 1.2232,
      "step": 382
    },
    {
      "epoch": 0.53,
      "learning_rate": 1e-05,
      "loss": 1.3826,
      "step": 383
    },
    {
      "epoch": 0.53,
      "learning_rate": 1e-05,
      "loss": 1.448,
      "step": 384
    },
    {
      "epoch": 0.53,
      "learning_rate": 1e-05,
      "loss": 1.3078,
      "step": 385
    },
    {
      "epoch": 0.53,
      "learning_rate": 1e-05,
      "loss": 1.1926,
      "step": 386
    },
    {
      "epoch": 0.53,
      "learning_rate": 1e-05,
      "loss": 1.4119,
      "step": 387
    },
    {
      "epoch": 0.53,
      "learning_rate": 1e-05,
      "loss": 1.2808,
      "step": 388
    },
    {
      "epoch": 0.53,
      "learning_rate": 1e-05,
      "loss": 1.2987,
      "step": 389
    },
    {
      "epoch": 0.53,
      "learning_rate": 1e-05,
      "loss": 1.2261,
      "step": 390
    },
    {
      "epoch": 0.54,
      "learning_rate": 1e-05,
      "loss": 1.3246,
      "step": 391
    },
    {
      "epoch": 0.54,
      "learning_rate": 1e-05,
      "loss": 1.4639,
      "step": 392
    },
    {
      "epoch": 0.54,
      "learning_rate": 1e-05,
      "loss": 1.2279,
      "step": 393
    },
    {
      "epoch": 0.54,
      "learning_rate": 1e-05,
      "loss": 1.53,
      "step": 394
    },
    {
      "epoch": 0.54,
      "learning_rate": 1e-05,
      "loss": 1.3108,
      "step": 395
    },
    {
      "epoch": 0.54,
      "learning_rate": 1e-05,
      "loss": 1.3875,
      "step": 396
    },
    {
      "epoch": 0.54,
      "learning_rate": 1e-05,
      "loss": 1.2812,
      "step": 397
    },
    {
      "epoch": 0.55,
      "learning_rate": 1e-05,
      "loss": 1.1143,
      "step": 398
    },
    {
      "epoch": 0.55,
      "learning_rate": 1e-05,
      "loss": 1.0317,
      "step": 399
    },
    {
      "epoch": 0.55,
      "learning_rate": 1e-05,
      "loss": 1.1922,
      "step": 400
    },
    {
      "epoch": 0.55,
      "learning_rate": 1e-05,
      "loss": 1.2728,
      "step": 401
    },
    {
      "epoch": 0.55,
      "learning_rate": 1e-05,
      "loss": 1.4347,
      "step": 402
    },
    {
      "epoch": 0.55,
      "learning_rate": 1e-05,
      "loss": 1.222,
      "step": 403
    },
    {
      "epoch": 0.55,
      "learning_rate": 1e-05,
      "loss": 1.3463,
      "step": 404
    },
    {
      "epoch": 0.56,
      "learning_rate": 1e-05,
      "loss": 1.3814,
      "step": 405
    },
    {
      "epoch": 0.56,
      "learning_rate": 1e-05,
      "loss": 1.0195,
      "step": 406
    },
    {
      "epoch": 0.56,
      "learning_rate": 1e-05,
      "loss": 1.3031,
      "step": 407
    },
    {
      "epoch": 0.56,
      "learning_rate": 1e-05,
      "loss": 1.1407,
      "step": 408
    },
    {
      "epoch": 0.56,
      "learning_rate": 1e-05,
      "loss": 1.3196,
      "step": 409
    },
    {
      "epoch": 0.56,
      "learning_rate": 1e-05,
      "loss": 1.5662,
      "step": 410
    },
    {
      "epoch": 0.56,
      "learning_rate": 1e-05,
      "loss": 1.4519,
      "step": 411
    },
    {
      "epoch": 0.56,
      "learning_rate": 1e-05,
      "loss": 1.271,
      "step": 412
    },
    {
      "epoch": 0.57,
      "learning_rate": 1e-05,
      "loss": 1.126,
      "step": 413
    },
    {
      "epoch": 0.57,
      "learning_rate": 1e-05,
      "loss": 1.2276,
      "step": 414
    },
    {
      "epoch": 0.57,
      "learning_rate": 1e-05,
      "loss": 1.4225,
      "step": 415
    },
    {
      "epoch": 0.57,
      "learning_rate": 1e-05,
      "loss": 1.4327,
      "step": 416
    },
    {
      "epoch": 0.57,
      "learning_rate": 1e-05,
      "loss": 1.2749,
      "step": 417
    },
    {
      "epoch": 0.57,
      "learning_rate": 1e-05,
      "loss": 1.537,
      "step": 418
    },
    {
      "epoch": 0.57,
      "learning_rate": 1e-05,
      "loss": 1.4696,
      "step": 419
    },
    {
      "epoch": 0.58,
      "learning_rate": 1e-05,
      "loss": 1.2686,
      "step": 420
    },
    {
      "epoch": 0.58,
      "learning_rate": 1e-05,
      "loss": 1.3604,
      "step": 421
    },
    {
      "epoch": 0.58,
      "learning_rate": 1e-05,
      "loss": 1.2854,
      "step": 422
    },
    {
      "epoch": 0.58,
      "learning_rate": 1e-05,
      "loss": 1.0215,
      "step": 423
    },
    {
      "epoch": 0.58,
      "learning_rate": 1e-05,
      "loss": 1.2871,
      "step": 424
    },
    {
      "epoch": 0.58,
      "learning_rate": 1e-05,
      "loss": 1.3841,
      "step": 425
    },
    {
      "epoch": 0.58,
      "learning_rate": 1e-05,
      "loss": 1.4487,
      "step": 426
    },
    {
      "epoch": 0.59,
      "learning_rate": 1e-05,
      "loss": 1.3908,
      "step": 427
    },
    {
      "epoch": 0.59,
      "learning_rate": 1e-05,
      "loss": 1.4015,
      "step": 428
    },
    {
      "epoch": 0.59,
      "learning_rate": 1e-05,
      "loss": 1.2585,
      "step": 429
    },
    {
      "epoch": 0.59,
      "learning_rate": 1e-05,
      "loss": 1.3472,
      "step": 430
    },
    {
      "epoch": 0.59,
      "learning_rate": 1e-05,
      "loss": 1.0587,
      "step": 431
    },
    {
      "epoch": 0.59,
      "learning_rate": 1e-05,
      "loss": 1.3713,
      "step": 432
    },
    {
      "epoch": 0.59,
      "learning_rate": 1e-05,
      "loss": 1.2885,
      "step": 433
    },
    {
      "epoch": 0.59,
      "learning_rate": 1e-05,
      "loss": 1.3062,
      "step": 434
    },
    {
      "epoch": 0.6,
      "learning_rate": 1e-05,
      "loss": 1.2972,
      "step": 435
    },
    {
      "epoch": 0.6,
      "learning_rate": 1e-05,
      "loss": 1.1511,
      "step": 436
    },
    {
      "epoch": 0.6,
      "learning_rate": 1e-05,
      "loss": 1.0984,
      "step": 437
    },
    {
      "epoch": 0.6,
      "learning_rate": 1e-05,
      "loss": 1.3867,
      "step": 438
    },
    {
      "epoch": 0.6,
      "learning_rate": 1e-05,
      "loss": 1.2911,
      "step": 439
    },
    {
      "epoch": 0.6,
      "learning_rate": 1e-05,
      "loss": 1.3166,
      "step": 440
    },
    {
      "epoch": 0.6,
      "learning_rate": 1e-05,
      "loss": 1.2634,
      "step": 441
    },
    {
      "epoch": 0.61,
      "learning_rate": 1e-05,
      "loss": 1.15,
      "step": 442
    },
    {
      "epoch": 0.61,
      "learning_rate": 1e-05,
      "loss": 1.4679,
      "step": 443
    },
    {
      "epoch": 0.61,
      "learning_rate": 1e-05,
      "loss": 1.0064,
      "step": 444
    },
    {
      "epoch": 0.61,
      "learning_rate": 1e-05,
      "loss": 1.2911,
      "step": 445
    },
    {
      "epoch": 0.61,
      "learning_rate": 1e-05,
      "loss": 1.3591,
      "step": 446
    },
    {
      "epoch": 0.61,
      "learning_rate": 1e-05,
      "loss": 1.1859,
      "step": 447
    },
    {
      "epoch": 0.61,
      "learning_rate": 1e-05,
      "loss": 1.0636,
      "step": 448
    },
    {
      "epoch": 0.62,
      "learning_rate": 1e-05,
      "loss": 1.3457,
      "step": 449
    },
    {
      "epoch": 0.62,
      "learning_rate": 1e-05,
      "loss": 1.1422,
      "step": 450
    },
    {
      "epoch": 0.62,
      "learning_rate": 1e-05,
      "loss": 1.4494,
      "step": 451
    },
    {
      "epoch": 0.62,
      "learning_rate": 1e-05,
      "loss": 1.196,
      "step": 452
    },
    {
      "epoch": 0.62,
      "learning_rate": 1e-05,
      "loss": 1.3039,
      "step": 453
    },
    {
      "epoch": 0.62,
      "learning_rate": 1e-05,
      "loss": 1.353,
      "step": 454
    },
    {
      "epoch": 0.62,
      "learning_rate": 1e-05,
      "loss": 1.3644,
      "step": 455
    },
    {
      "epoch": 0.63,
      "learning_rate": 1e-05,
      "loss": 1.1362,
      "step": 456
    },
    {
      "epoch": 0.63,
      "learning_rate": 1e-05,
      "loss": 1.2784,
      "step": 457
    },
    {
      "epoch": 0.63,
      "learning_rate": 1e-05,
      "loss": 1.3314,
      "step": 458
    },
    {
      "epoch": 0.63,
      "learning_rate": 1e-05,
      "loss": 1.3801,
      "step": 459
    },
    {
      "epoch": 0.63,
      "learning_rate": 1e-05,
      "loss": 1.2218,
      "step": 460
    },
    {
      "epoch": 0.63,
      "learning_rate": 1e-05,
      "loss": 1.021,
      "step": 461
    },
    {
      "epoch": 0.63,
      "learning_rate": 1e-05,
      "loss": 1.0561,
      "step": 462
    },
    {
      "epoch": 0.63,
      "learning_rate": 1e-05,
      "loss": 1.0826,
      "step": 463
    },
    {
      "epoch": 0.64,
      "learning_rate": 1e-05,
      "loss": 1.19,
      "step": 464
    },
    {
      "epoch": 0.64,
      "learning_rate": 1e-05,
      "loss": 1.4989,
      "step": 465
    },
    {
      "epoch": 0.64,
      "learning_rate": 1e-05,
      "loss": 1.2837,
      "step": 466
    },
    {
      "epoch": 0.64,
      "learning_rate": 1e-05,
      "loss": 1.3075,
      "step": 467
    },
    {
      "epoch": 0.64,
      "learning_rate": 1e-05,
      "loss": 1.1891,
      "step": 468
    },
    {
      "epoch": 0.64,
      "learning_rate": 1e-05,
      "loss": 1.3762,
      "step": 469
    },
    {
      "epoch": 0.64,
      "learning_rate": 1e-05,
      "loss": 1.295,
      "step": 470
    },
    {
      "epoch": 0.65,
      "learning_rate": 1e-05,
      "loss": 1.2485,
      "step": 471
    },
    {
      "epoch": 0.65,
      "learning_rate": 1e-05,
      "loss": 1.4234,
      "step": 472
    },
    {
      "epoch": 0.65,
      "learning_rate": 1e-05,
      "loss": 1.0153,
      "step": 473
    },
    {
      "epoch": 0.65,
      "learning_rate": 1e-05,
      "loss": 1.367,
      "step": 474
    },
    {
      "epoch": 0.65,
      "learning_rate": 1e-05,
      "loss": 1.1507,
      "step": 475
    },
    {
      "epoch": 0.65,
      "learning_rate": 1e-05,
      "loss": 1.4302,
      "step": 476
    },
    {
      "epoch": 0.65,
      "learning_rate": 1e-05,
      "loss": 1.3279,
      "step": 477
    },
    {
      "epoch": 0.66,
      "learning_rate": 1e-05,
      "loss": 1.2813,
      "step": 478
    },
    {
      "epoch": 0.66,
      "learning_rate": 1e-05,
      "loss": 1.1196,
      "step": 479
    },
    {
      "epoch": 0.66,
      "learning_rate": 1e-05,
      "loss": 1.356,
      "step": 480
    },
    {
      "epoch": 0.66,
      "learning_rate": 1e-05,
      "loss": 0.9829,
      "step": 481
    },
    {
      "epoch": 0.66,
      "learning_rate": 1e-05,
      "loss": 1.1147,
      "step": 482
    },
    {
      "epoch": 0.66,
      "learning_rate": 1e-05,
      "loss": 1.1877,
      "step": 483
    },
    {
      "epoch": 0.66,
      "learning_rate": 1e-05,
      "loss": 1.1869,
      "step": 484
    },
    {
      "epoch": 0.66,
      "learning_rate": 1e-05,
      "loss": 1.2238,
      "step": 485
    },
    {
      "epoch": 0.67,
      "learning_rate": 1e-05,
      "loss": 1.0242,
      "step": 486
    },
    {
      "epoch": 0.67,
      "learning_rate": 1e-05,
      "loss": 1.3016,
      "step": 487
    },
    {
      "epoch": 0.67,
      "learning_rate": 1e-05,
      "loss": 1.2528,
      "step": 488
    },
    {
      "epoch": 0.67,
      "learning_rate": 1e-05,
      "loss": 1.4698,
      "step": 489
    },
    {
      "epoch": 0.67,
      "learning_rate": 1e-05,
      "loss": 1.2011,
      "step": 490
    },
    {
      "epoch": 0.67,
      "learning_rate": 1e-05,
      "loss": 1.298,
      "step": 491
    },
    {
      "epoch": 0.67,
      "learning_rate": 1e-05,
      "loss": 0.977,
      "step": 492
    },
    {
      "epoch": 0.68,
      "learning_rate": 1e-05,
      "loss": 1.3224,
      "step": 493
    },
    {
      "epoch": 0.68,
      "learning_rate": 1e-05,
      "loss": 1.1522,
      "step": 494
    },
    {
      "epoch": 0.68,
      "learning_rate": 1e-05,
      "loss": 1.3151,
      "step": 495
    },
    {
      "epoch": 0.68,
      "learning_rate": 1e-05,
      "loss": 0.9746,
      "step": 496
    },
    {
      "epoch": 0.68,
      "learning_rate": 1e-05,
      "loss": 1.167,
      "step": 497
    },
    {
      "epoch": 0.68,
      "learning_rate": 1e-05,
      "loss": 1.2231,
      "step": 498
    },
    {
      "epoch": 0.68,
      "learning_rate": 1e-05,
      "loss": 1.2048,
      "step": 499
    },
    {
      "epoch": 0.69,
      "learning_rate": 1e-05,
      "loss": 0.9615,
      "step": 500
    },
    {
      "epoch": 0.69,
      "learning_rate": 1e-05,
      "loss": 0.9769,
      "step": 501
    },
    {
      "epoch": 0.69,
      "learning_rate": 1e-05,
      "loss": 1.2813,
      "step": 502
    },
    {
      "epoch": 0.69,
      "learning_rate": 1e-05,
      "loss": 1.2979,
      "step": 503
    },
    {
      "epoch": 0.69,
      "learning_rate": 1e-05,
      "loss": 1.1847,
      "step": 504
    },
    {
      "epoch": 0.69,
      "learning_rate": 1e-05,
      "loss": 1.2916,
      "step": 505
    },
    {
      "epoch": 0.69,
      "learning_rate": 1e-05,
      "loss": 1.2963,
      "step": 506
    },
    {
      "epoch": 0.69,
      "learning_rate": 1e-05,
      "loss": 1.2797,
      "step": 507
    },
    {
      "epoch": 0.7,
      "learning_rate": 1e-05,
      "loss": 1.5743,
      "step": 508
    },
    {
      "epoch": 0.7,
      "learning_rate": 1e-05,
      "loss": 1.2985,
      "step": 509
    },
    {
      "epoch": 0.7,
      "learning_rate": 1e-05,
      "loss": 1.3771,
      "step": 510
    },
    {
      "epoch": 0.7,
      "learning_rate": 1e-05,
      "loss": 1.2798,
      "step": 511
    },
    {
      "epoch": 0.7,
      "learning_rate": 1e-05,
      "loss": 1.0678,
      "step": 512
    },
    {
      "epoch": 0.7,
      "learning_rate": 1e-05,
      "loss": 1.3013,
      "step": 513
    },
    {
      "epoch": 0.7,
      "learning_rate": 1e-05,
      "loss": 1.2849,
      "step": 514
    },
    {
      "epoch": 0.71,
      "learning_rate": 1e-05,
      "loss": 1.1557,
      "step": 515
    },
    {
      "epoch": 0.71,
      "learning_rate": 1e-05,
      "loss": 1.353,
      "step": 516
    },
    {
      "epoch": 0.71,
      "learning_rate": 1e-05,
      "loss": 1.3528,
      "step": 517
    },
    {
      "epoch": 0.71,
      "learning_rate": 1e-05,
      "loss": 1.2992,
      "step": 518
    },
    {
      "epoch": 0.71,
      "learning_rate": 1e-05,
      "loss": 1.3185,
      "step": 519
    },
    {
      "epoch": 0.71,
      "learning_rate": 1e-05,
      "loss": 1.1575,
      "step": 520
    },
    {
      "epoch": 0.71,
      "learning_rate": 1e-05,
      "loss": 1.1214,
      "step": 521
    },
    {
      "epoch": 0.72,
      "learning_rate": 1e-05,
      "loss": 0.9405,
      "step": 522
    },
    {
      "epoch": 0.72,
      "learning_rate": 1e-05,
      "loss": 1.3758,
      "step": 523
    },
    {
      "epoch": 0.72,
      "learning_rate": 1e-05,
      "loss": 1.3093,
      "step": 524
    },
    {
      "epoch": 0.72,
      "learning_rate": 1e-05,
      "loss": 1.2981,
      "step": 525
    },
    {
      "epoch": 0.72,
      "learning_rate": 1e-05,
      "loss": 1.1522,
      "step": 526
    },
    {
      "epoch": 0.72,
      "learning_rate": 1e-05,
      "loss": 1.322,
      "step": 527
    },
    {
      "epoch": 0.72,
      "learning_rate": 1e-05,
      "loss": 1.2559,
      "step": 528
    },
    {
      "epoch": 0.73,
      "learning_rate": 1e-05,
      "loss": 1.1047,
      "step": 529
    },
    {
      "epoch": 0.73,
      "learning_rate": 1e-05,
      "loss": 1.4091,
      "step": 530
    },
    {
      "epoch": 0.73,
      "learning_rate": 1e-05,
      "loss": 1.0045,
      "step": 531
    },
    {
      "epoch": 0.73,
      "learning_rate": 1e-05,
      "loss": 1.2765,
      "step": 532
    },
    {
      "epoch": 0.73,
      "learning_rate": 1e-05,
      "loss": 1.2671,
      "step": 533
    },
    {
      "epoch": 0.73,
      "learning_rate": 1e-05,
      "loss": 1.2466,
      "step": 534
    },
    {
      "epoch": 0.73,
      "learning_rate": 1e-05,
      "loss": 1.3524,
      "step": 535
    },
    {
      "epoch": 0.73,
      "learning_rate": 1e-05,
      "loss": 1.2466,
      "step": 536
    },
    {
      "epoch": 0.74,
      "learning_rate": 1e-05,
      "loss": 1.1754,
      "step": 537
    },
    {
      "epoch": 0.74,
      "learning_rate": 1e-05,
      "loss": 1.4108,
      "step": 538
    },
    {
      "epoch": 0.74,
      "learning_rate": 1e-05,
      "loss": 1.0907,
      "step": 539
    },
    {
      "epoch": 0.74,
      "learning_rate": 1e-05,
      "loss": 1.2495,
      "step": 540
    },
    {
      "epoch": 0.74,
      "learning_rate": 1e-05,
      "loss": 1.4228,
      "step": 541
    },
    {
      "epoch": 0.74,
      "learning_rate": 1e-05,
      "loss": 1.3303,
      "step": 542
    },
    {
      "epoch": 0.74,
      "learning_rate": 1e-05,
      "loss": 1.4513,
      "step": 543
    },
    {
      "epoch": 0.75,
      "learning_rate": 1e-05,
      "loss": 1.0478,
      "step": 544
    },
    {
      "epoch": 0.75,
      "learning_rate": 1e-05,
      "loss": 1.1141,
      "step": 545
    },
    {
      "epoch": 0.75,
      "learning_rate": 1e-05,
      "loss": 1.1941,
      "step": 546
    },
    {
      "epoch": 0.75,
      "learning_rate": 1e-05,
      "loss": 1.2352,
      "step": 547
    },
    {
      "epoch": 0.75,
      "learning_rate": 1e-05,
      "loss": 1.3497,
      "step": 548
    },
    {
      "epoch": 0.75,
      "learning_rate": 1e-05,
      "loss": 1.276,
      "step": 549
    },
    {
      "epoch": 0.75,
      "learning_rate": 1e-05,
      "loss": 1.2045,
      "step": 550
    },
    {
      "epoch": 0.76,
      "learning_rate": 1e-05,
      "loss": 1.0945,
      "step": 551
    },
    {
      "epoch": 0.76,
      "learning_rate": 1e-05,
      "loss": 1.1572,
      "step": 552
    },
    {
      "epoch": 0.76,
      "learning_rate": 1e-05,
      "loss": 1.4123,
      "step": 553
    },
    {
      "epoch": 0.76,
      "learning_rate": 1e-05,
      "loss": 1.2647,
      "step": 554
    },
    {
      "epoch": 0.76,
      "learning_rate": 1e-05,
      "loss": 1.1335,
      "step": 555
    },
    {
      "epoch": 0.76,
      "learning_rate": 1e-05,
      "loss": 1.2252,
      "step": 556
    },
    {
      "epoch": 0.76,
      "learning_rate": 1e-05,
      "loss": 1.1607,
      "step": 557
    },
    {
      "epoch": 0.76,
      "learning_rate": 1e-05,
      "loss": 1.2649,
      "step": 558
    },
    {
      "epoch": 0.77,
      "learning_rate": 1e-05,
      "loss": 1.1959,
      "step": 559
    },
    {
      "epoch": 0.77,
      "learning_rate": 1e-05,
      "loss": 1.3358,
      "step": 560
    },
    {
      "epoch": 0.77,
      "learning_rate": 1e-05,
      "loss": 1.2121,
      "step": 561
    },
    {
      "epoch": 0.77,
      "learning_rate": 1e-05,
      "loss": 1.1291,
      "step": 562
    },
    {
      "epoch": 0.77,
      "learning_rate": 1e-05,
      "loss": 1.1961,
      "step": 563
    },
    {
      "epoch": 0.77,
      "learning_rate": 1e-05,
      "loss": 1.2156,
      "step": 564
    },
    {
      "epoch": 0.77,
      "learning_rate": 1e-05,
      "loss": 1.1001,
      "step": 565
    },
    {
      "epoch": 0.78,
      "learning_rate": 1e-05,
      "loss": 1.2442,
      "step": 566
    },
    {
      "epoch": 0.78,
      "learning_rate": 1e-05,
      "loss": 1.3107,
      "step": 567
    },
    {
      "epoch": 0.78,
      "learning_rate": 1e-05,
      "loss": 1.1548,
      "step": 568
    },
    {
      "epoch": 0.78,
      "learning_rate": 1e-05,
      "loss": 1.3009,
      "step": 569
    },
    {
      "epoch": 0.78,
      "learning_rate": 1e-05,
      "loss": 1.2116,
      "step": 570
    },
    {
      "epoch": 0.78,
      "learning_rate": 1e-05,
      "loss": 1.1632,
      "step": 571
    },
    {
      "epoch": 0.78,
      "learning_rate": 1e-05,
      "loss": 1.2011,
      "step": 572
    },
    {
      "epoch": 0.79,
      "learning_rate": 1e-05,
      "loss": 0.9512,
      "step": 573
    },
    {
      "epoch": 0.79,
      "learning_rate": 1e-05,
      "loss": 1.1912,
      "step": 574
    },
    {
      "epoch": 0.79,
      "learning_rate": 1e-05,
      "loss": 1.0888,
      "step": 575
    },
    {
      "epoch": 0.79,
      "learning_rate": 1e-05,
      "loss": 1.2705,
      "step": 576
    },
    {
      "epoch": 0.79,
      "learning_rate": 1e-05,
      "loss": 1.1161,
      "step": 577
    },
    {
      "epoch": 0.79,
      "learning_rate": 1e-05,
      "loss": 1.1722,
      "step": 578
    },
    {
      "epoch": 0.79,
      "learning_rate": 1e-05,
      "loss": 1.3802,
      "step": 579
    },
    {
      "epoch": 0.8,
      "learning_rate": 1e-05,
      "loss": 1.0887,
      "step": 580
    },
    {
      "epoch": 0.8,
      "learning_rate": 1e-05,
      "loss": 1.268,
      "step": 581
    },
    {
      "epoch": 0.8,
      "learning_rate": 1e-05,
      "loss": 0.8627,
      "step": 582
    },
    {
      "epoch": 0.8,
      "learning_rate": 1e-05,
      "loss": 1.2565,
      "step": 583
    },
    {
      "epoch": 0.8,
      "learning_rate": 1e-05,
      "loss": 1.1433,
      "step": 584
    },
    {
      "epoch": 0.8,
      "learning_rate": 1e-05,
      "loss": 1.2263,
      "step": 585
    },
    {
      "epoch": 0.8,
      "learning_rate": 1e-05,
      "loss": 1.1606,
      "step": 586
    },
    {
      "epoch": 0.8,
      "learning_rate": 1e-05,
      "loss": 1.2557,
      "step": 587
    },
    {
      "epoch": 0.81,
      "learning_rate": 1e-05,
      "loss": 1.2018,
      "step": 588
    },
    {
      "epoch": 0.81,
      "learning_rate": 1e-05,
      "loss": 1.1545,
      "step": 589
    },
    {
      "epoch": 0.81,
      "learning_rate": 1e-05,
      "loss": 1.0917,
      "step": 590
    },
    {
      "epoch": 0.81,
      "learning_rate": 1e-05,
      "loss": 1.1935,
      "step": 591
    },
    {
      "epoch": 0.81,
      "learning_rate": 1e-05,
      "loss": 1.1721,
      "step": 592
    },
    {
      "epoch": 0.81,
      "learning_rate": 1e-05,
      "loss": 1.4243,
      "step": 593
    },
    {
      "epoch": 0.81,
      "learning_rate": 1e-05,
      "loss": 0.9393,
      "step": 594
    },
    {
      "epoch": 0.82,
      "learning_rate": 1e-05,
      "loss": 1.2742,
      "step": 595
    },
    {
      "epoch": 0.82,
      "learning_rate": 1e-05,
      "loss": 1.0141,
      "step": 596
    },
    {
      "epoch": 0.82,
      "learning_rate": 1e-05,
      "loss": 0.9228,
      "step": 597
    },
    {
      "epoch": 0.82,
      "learning_rate": 1e-05,
      "loss": 1.2762,
      "step": 598
    },
    {
      "epoch": 0.82,
      "learning_rate": 1e-05,
      "loss": 1.1166,
      "step": 599
    },
    {
      "epoch": 0.82,
      "learning_rate": 1e-05,
      "loss": 0.9603,
      "step": 600
    },
    {
      "epoch": 0.82,
      "learning_rate": 1e-05,
      "loss": 1.1453,
      "step": 601
    },
    {
      "epoch": 0.83,
      "learning_rate": 1e-05,
      "loss": 1.0849,
      "step": 602
    },
    {
      "epoch": 0.83,
      "learning_rate": 1e-05,
      "loss": 1.288,
      "step": 603
    },
    {
      "epoch": 0.83,
      "learning_rate": 1e-05,
      "loss": 1.0641,
      "step": 604
    },
    {
      "epoch": 0.83,
      "learning_rate": 1e-05,
      "loss": 1.2234,
      "step": 605
    },
    {
      "epoch": 0.83,
      "learning_rate": 1e-05,
      "loss": 1.2774,
      "step": 606
    },
    {
      "epoch": 0.83,
      "learning_rate": 1e-05,
      "loss": 1.1161,
      "step": 607
    },
    {
      "epoch": 0.83,
      "learning_rate": 1e-05,
      "loss": 1.4146,
      "step": 608
    },
    {
      "epoch": 0.83,
      "learning_rate": 1e-05,
      "loss": 1.1844,
      "step": 609
    },
    {
      "epoch": 0.84,
      "learning_rate": 1e-05,
      "loss": 1.3707,
      "step": 610
    },
    {
      "epoch": 0.84,
      "learning_rate": 1e-05,
      "loss": 1.3002,
      "step": 611
    },
    {
      "epoch": 0.84,
      "learning_rate": 1e-05,
      "loss": 1.2849,
      "step": 612
    },
    {
      "epoch": 0.84,
      "learning_rate": 1e-05,
      "loss": 1.0489,
      "step": 613
    },
    {
      "epoch": 0.84,
      "learning_rate": 1e-05,
      "loss": 1.1897,
      "step": 614
    },
    {
      "epoch": 0.84,
      "learning_rate": 1e-05,
      "loss": 1.09,
      "step": 615
    },
    {
      "epoch": 0.84,
      "learning_rate": 1e-05,
      "loss": 1.1779,
      "step": 616
    },
    {
      "epoch": 0.85,
      "learning_rate": 1e-05,
      "loss": 1.3649,
      "step": 617
    },
    {
      "epoch": 0.85,
      "learning_rate": 1e-05,
      "loss": 1.1049,
      "step": 618
    },
    {
      "epoch": 0.85,
      "learning_rate": 1e-05,
      "loss": 1.257,
      "step": 619
    },
    {
      "epoch": 0.85,
      "learning_rate": 1e-05,
      "loss": 1.273,
      "step": 620
    },
    {
      "epoch": 0.85,
      "learning_rate": 1e-05,
      "loss": 0.8941,
      "step": 621
    },
    {
      "epoch": 0.85,
      "learning_rate": 1e-05,
      "loss": 1.1427,
      "step": 622
    },
    {
      "epoch": 0.85,
      "learning_rate": 1e-05,
      "loss": 1.1603,
      "step": 623
    },
    {
      "epoch": 0.86,
      "learning_rate": 1e-05,
      "loss": 1.1058,
      "step": 624
    },
    {
      "epoch": 0.86,
      "learning_rate": 1e-05,
      "loss": 1.136,
      "step": 625
    },
    {
      "epoch": 0.86,
      "learning_rate": 1e-05,
      "loss": 1.3001,
      "step": 626
    },
    {
      "epoch": 0.86,
      "learning_rate": 1e-05,
      "loss": 0.8937,
      "step": 627
    },
    {
      "epoch": 0.86,
      "learning_rate": 1e-05,
      "loss": 1.2081,
      "step": 628
    },
    {
      "epoch": 0.86,
      "learning_rate": 1e-05,
      "loss": 1.3189,
      "step": 629
    },
    {
      "epoch": 0.86,
      "learning_rate": 1e-05,
      "loss": 1.082,
      "step": 630
    },
    {
      "epoch": 0.86,
      "learning_rate": 1e-05,
      "loss": 1.0969,
      "step": 631
    },
    {
      "epoch": 0.87,
      "learning_rate": 1e-05,
      "loss": 1.2862,
      "step": 632
    },
    {
      "epoch": 0.87,
      "learning_rate": 1e-05,
      "loss": 1.0158,
      "step": 633
    },
    {
      "epoch": 0.87,
      "learning_rate": 1e-05,
      "loss": 1.1292,
      "step": 634
    },
    {
      "epoch": 0.87,
      "learning_rate": 1e-05,
      "loss": 1.1611,
      "step": 635
    },
    {
      "epoch": 0.87,
      "learning_rate": 1e-05,
      "loss": 1.0885,
      "step": 636
    },
    {
      "epoch": 0.87,
      "learning_rate": 1e-05,
      "loss": 1.0353,
      "step": 637
    },
    {
      "epoch": 0.87,
      "learning_rate": 1e-05,
      "loss": 1.5197,
      "step": 638
    },
    {
      "epoch": 0.88,
      "learning_rate": 1e-05,
      "loss": 1.2418,
      "step": 639
    },
    {
      "epoch": 0.88,
      "learning_rate": 1e-05,
      "loss": 1.0245,
      "step": 640
    },
    {
      "epoch": 0.88,
      "learning_rate": 1e-05,
      "loss": 0.9594,
      "step": 641
    },
    {
      "epoch": 0.88,
      "learning_rate": 1e-05,
      "loss": 1.2253,
      "step": 642
    },
    {
      "epoch": 0.88,
      "learning_rate": 1e-05,
      "loss": 1.1019,
      "step": 643
    },
    {
      "epoch": 0.88,
      "learning_rate": 1e-05,
      "loss": 1.0199,
      "step": 644
    },
    {
      "epoch": 0.88,
      "learning_rate": 1e-05,
      "loss": 0.9892,
      "step": 645
    },
    {
      "epoch": 0.89,
      "learning_rate": 1e-05,
      "loss": 1.0112,
      "step": 646
    },
    {
      "epoch": 0.89,
      "learning_rate": 1e-05,
      "loss": 0.9136,
      "step": 647
    },
    {
      "epoch": 0.89,
      "learning_rate": 1e-05,
      "loss": 1.1581,
      "step": 648
    },
    {
      "epoch": 0.89,
      "learning_rate": 1e-05,
      "loss": 1.1571,
      "step": 649
    },
    {
      "epoch": 0.89,
      "learning_rate": 1e-05,
      "loss": 1.5382,
      "step": 650
    },
    {
      "epoch": 0.89,
      "learning_rate": 1e-05,
      "loss": 1.0162,
      "step": 651
    },
    {
      "epoch": 0.89,
      "learning_rate": 1e-05,
      "loss": 1.1514,
      "step": 652
    },
    {
      "epoch": 0.9,
      "learning_rate": 1e-05,
      "loss": 1.2793,
      "step": 653
    },
    {
      "epoch": 0.9,
      "learning_rate": 1e-05,
      "loss": 1.1468,
      "step": 654
    },
    {
      "epoch": 0.9,
      "learning_rate": 1e-05,
      "loss": 1.4307,
      "step": 655
    },
    {
      "epoch": 0.9,
      "learning_rate": 1e-05,
      "loss": 1.0331,
      "step": 656
    },
    {
      "epoch": 0.9,
      "learning_rate": 1e-05,
      "loss": 1.3126,
      "step": 657
    },
    {
      "epoch": 0.9,
      "learning_rate": 1e-05,
      "loss": 1.2361,
      "step": 658
    },
    {
      "epoch": 0.9,
      "learning_rate": 1e-05,
      "loss": 1.1493,
      "step": 659
    },
    {
      "epoch": 0.9,
      "learning_rate": 1e-05,
      "loss": 0.9919,
      "step": 660
    },
    {
      "epoch": 0.91,
      "learning_rate": 1e-05,
      "loss": 1.3145,
      "step": 661
    },
    {
      "epoch": 0.91,
      "learning_rate": 1e-05,
      "loss": 1.0194,
      "step": 662
    },
    {
      "epoch": 0.91,
      "learning_rate": 1e-05,
      "loss": 1.2119,
      "step": 663
    },
    {
      "epoch": 0.91,
      "learning_rate": 1e-05,
      "loss": 1.4123,
      "step": 664
    },
    {
      "epoch": 0.91,
      "learning_rate": 1e-05,
      "loss": 1.4708,
      "step": 665
    },
    {
      "epoch": 0.91,
      "learning_rate": 1e-05,
      "loss": 0.9285,
      "step": 666
    },
    {
      "epoch": 0.91,
      "learning_rate": 1e-05,
      "loss": 1.2336,
      "step": 667
    },
    {
      "epoch": 0.92,
      "learning_rate": 1e-05,
      "loss": 1.2068,
      "step": 668
    },
    {
      "epoch": 0.92,
      "learning_rate": 1e-05,
      "loss": 1.1762,
      "step": 669
    },
    {
      "epoch": 0.92,
      "learning_rate": 1e-05,
      "loss": 1.0675,
      "step": 670
    },
    {
      "epoch": 0.92,
      "learning_rate": 1e-05,
      "loss": 1.1717,
      "step": 671
    },
    {
      "epoch": 0.92,
      "learning_rate": 1e-05,
      "loss": 1.2715,
      "step": 672
    },
    {
      "epoch": 0.92,
      "learning_rate": 1e-05,
      "loss": 1.2124,
      "step": 673
    },
    {
      "epoch": 0.92,
      "learning_rate": 1e-05,
      "loss": 0.9171,
      "step": 674
    },
    {
      "epoch": 0.93,
      "learning_rate": 1e-05,
      "loss": 1.1485,
      "step": 675
    },
    {
      "epoch": 0.93,
      "learning_rate": 1e-05,
      "loss": 1.065,
      "step": 676
    },
    {
      "epoch": 0.93,
      "learning_rate": 1e-05,
      "loss": 0.9832,
      "step": 677
    },
    {
      "epoch": 0.93,
      "learning_rate": 1e-05,
      "loss": 1.1978,
      "step": 678
    },
    {
      "epoch": 0.93,
      "learning_rate": 1e-05,
      "loss": 1.2763,
      "step": 679
    },
    {
      "epoch": 0.93,
      "learning_rate": 1e-05,
      "loss": 1.3615,
      "step": 680
    },
    {
      "epoch": 0.93,
      "learning_rate": 1e-05,
      "loss": 1.0458,
      "step": 681
    },
    {
      "epoch": 0.93,
      "learning_rate": 1e-05,
      "loss": 1.0635,
      "step": 682
    },
    {
      "epoch": 0.94,
      "learning_rate": 1e-05,
      "loss": 1.1269,
      "step": 683
    },
    {
      "epoch": 0.94,
      "learning_rate": 1e-05,
      "loss": 1.0314,
      "step": 684
    },
    {
      "epoch": 0.94,
      "learning_rate": 1e-05,
      "loss": 1.0545,
      "step": 685
    },
    {
      "epoch": 0.94,
      "learning_rate": 1e-05,
      "loss": 1.2244,
      "step": 686
    },
    {
      "epoch": 0.94,
      "learning_rate": 1e-05,
      "loss": 1.2855,
      "step": 687
    },
    {
      "epoch": 0.94,
      "learning_rate": 1e-05,
      "loss": 1.2325,
      "step": 688
    },
    {
      "epoch": 0.94,
      "learning_rate": 1e-05,
      "loss": 0.7963,
      "step": 689
    },
    {
      "epoch": 0.95,
      "learning_rate": 1e-05,
      "loss": 0.9491,
      "step": 690
    },
    {
      "epoch": 0.95,
      "learning_rate": 1e-05,
      "loss": 1.0848,
      "step": 691
    },
    {
      "epoch": 0.95,
      "learning_rate": 1e-05,
      "loss": 1.0316,
      "step": 692
    },
    {
      "epoch": 0.95,
      "learning_rate": 1e-05,
      "loss": 1.2156,
      "step": 693
    },
    {
      "epoch": 0.95,
      "learning_rate": 1e-05,
      "loss": 0.9738,
      "step": 694
    },
    {
      "epoch": 0.95,
      "learning_rate": 1e-05,
      "loss": 1.2488,
      "step": 695
    },
    {
      "epoch": 0.95,
      "learning_rate": 1e-05,
      "loss": 1.0594,
      "step": 696
    },
    {
      "epoch": 0.96,
      "learning_rate": 1e-05,
      "loss": 1.2164,
      "step": 697
    },
    {
      "epoch": 0.96,
      "learning_rate": 1e-05,
      "loss": 0.9978,
      "step": 698
    },
    {
      "epoch": 0.96,
      "learning_rate": 1e-05,
      "loss": 1.226,
      "step": 699
    },
    {
      "epoch": 0.96,
      "learning_rate": 1e-05,
      "loss": 1.1203,
      "step": 700
    },
    {
      "epoch": 0.96,
      "learning_rate": 1e-05,
      "loss": 1.1197,
      "step": 701
    },
    {
      "epoch": 0.96,
      "learning_rate": 1e-05,
      "loss": 1.1775,
      "step": 702
    },
    {
      "epoch": 0.96,
      "learning_rate": 1e-05,
      "loss": 1.2717,
      "step": 703
    },
    {
      "epoch": 0.97,
      "learning_rate": 1e-05,
      "loss": 1.0475,
      "step": 704
    },
    {
      "epoch": 0.97,
      "learning_rate": 1e-05,
      "loss": 0.959,
      "step": 705
    },
    {
      "epoch": 0.97,
      "learning_rate": 1e-05,
      "loss": 1.2826,
      "step": 706
    },
    {
      "epoch": 0.97,
      "learning_rate": 1e-05,
      "loss": 1.2405,
      "step": 707
    },
    {
      "epoch": 0.97,
      "learning_rate": 1e-05,
      "loss": 1.3496,
      "step": 708
    },
    {
      "epoch": 0.97,
      "learning_rate": 1e-05,
      "loss": 1.0772,
      "step": 709
    },
    {
      "epoch": 0.97,
      "learning_rate": 1e-05,
      "loss": 1.2431,
      "step": 710
    },
    {
      "epoch": 0.97,
      "learning_rate": 1e-05,
      "loss": 1.0681,
      "step": 711
    },
    {
      "epoch": 0.98,
      "learning_rate": 1e-05,
      "loss": 1.4682,
      "step": 712
    },
    {
      "epoch": 0.98,
      "learning_rate": 1e-05,
      "loss": 1.1159,
      "step": 713
    },
    {
      "epoch": 0.98,
      "learning_rate": 1e-05,
      "loss": 1.0574,
      "step": 714
    },
    {
      "epoch": 0.98,
      "learning_rate": 1e-05,
      "loss": 1.0319,
      "step": 715
    },
    {
      "epoch": 0.98,
      "learning_rate": 1e-05,
      "loss": 0.9838,
      "step": 716
    },
    {
      "epoch": 0.98,
      "learning_rate": 1e-05,
      "loss": 1.5624,
      "step": 717
    },
    {
      "epoch": 0.98,
      "learning_rate": 1e-05,
      "loss": 0.9486,
      "step": 718
    },
    {
      "epoch": 0.99,
      "learning_rate": 1e-05,
      "loss": 0.9797,
      "step": 719
    },
    {
      "epoch": 0.99,
      "learning_rate": 1e-05,
      "loss": 1.35,
      "step": 720
    },
    {
      "epoch": 0.99,
      "learning_rate": 1e-05,
      "loss": 1.0952,
      "step": 721
    },
    {
      "epoch": 0.99,
      "learning_rate": 1e-05,
      "loss": 1.1613,
      "step": 722
    },
    {
      "epoch": 0.99,
      "learning_rate": 1e-05,
      "loss": 1.2871,
      "step": 723
    },
    {
      "epoch": 0.99,
      "learning_rate": 1e-05,
      "loss": 1.1193,
      "step": 724
    },
    {
      "epoch": 0.99,
      "learning_rate": 1e-05,
      "loss": 1.1311,
      "step": 725
    },
    {
      "epoch": 1.0,
      "learning_rate": 1e-05,
      "loss": 1.1926,
      "step": 726
    },
    {
      "epoch": 1.0,
      "learning_rate": 1e-05,
      "loss": 1.4396,
      "step": 727
    },
    {
      "epoch": 1.0,
      "learning_rate": 1e-05,
      "loss": 1.0607,
      "step": 728
    },
    {
      "epoch": 1.0,
      "learning_rate": 1e-05,
      "loss": 1.2212,
      "step": 729
    },
    {
      "epoch": 1.0,
      "step": 729,
      "total_flos": 17757294821376.0,
      "train_loss": 1.4406553773395974,
      "train_runtime": 51387.1617,
      "train_samples_per_second": 0.454,
      "train_steps_per_second": 0.014
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 729,
  "num_train_epochs": 1,
  "save_steps": 1000,
  "total_flos": 17757294821376.0,
  "trial_name": null,
  "trial_params": null
}