{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.9977695167286247,
  "eval_steps": 500,
  "global_step": 756,
  "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": 1.5239,
      "step": 1
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.333333333333334e-06,
      "loss": 1.4583,
      "step": 2
    },
    {
      "epoch": 0.01,
      "learning_rate": 5.283208335737189e-06,
      "loss": 1.4205,
      "step": 3
    },
    {
      "epoch": 0.02,
      "learning_rate": 6.666666666666668e-06,
      "loss": 1.4054,
      "step": 4
    },
    {
      "epoch": 0.02,
      "learning_rate": 7.739760316291209e-06,
      "loss": 1.3651,
      "step": 5
    },
    {
      "epoch": 0.02,
      "learning_rate": 8.616541669070522e-06,
      "loss": 1.3724,
      "step": 6
    },
    {
      "epoch": 0.03,
      "learning_rate": 9.357849740192014e-06,
      "loss": 1.3489,
      "step": 7
    },
    {
      "epoch": 0.03,
      "learning_rate": 1e-05,
      "loss": 1.3221,
      "step": 8
    },
    {
      "epoch": 0.04,
      "learning_rate": 1e-05,
      "loss": 1.3005,
      "step": 9
    },
    {
      "epoch": 0.04,
      "learning_rate": 1e-05,
      "loss": 1.248,
      "step": 10
    },
    {
      "epoch": 0.04,
      "learning_rate": 1e-05,
      "loss": 1.2617,
      "step": 11
    },
    {
      "epoch": 0.05,
      "learning_rate": 1e-05,
      "loss": 1.2399,
      "step": 12
    },
    {
      "epoch": 0.05,
      "learning_rate": 1e-05,
      "loss": 1.2312,
      "step": 13
    },
    {
      "epoch": 0.06,
      "learning_rate": 1e-05,
      "loss": 1.2318,
      "step": 14
    },
    {
      "epoch": 0.06,
      "learning_rate": 1e-05,
      "loss": 1.2094,
      "step": 15
    },
    {
      "epoch": 0.06,
      "learning_rate": 1e-05,
      "loss": 1.2108,
      "step": 16
    },
    {
      "epoch": 0.07,
      "learning_rate": 1e-05,
      "loss": 1.218,
      "step": 17
    },
    {
      "epoch": 0.07,
      "learning_rate": 1e-05,
      "loss": 1.2062,
      "step": 18
    },
    {
      "epoch": 0.08,
      "learning_rate": 1e-05,
      "loss": 1.1602,
      "step": 19
    },
    {
      "epoch": 0.08,
      "learning_rate": 1e-05,
      "loss": 1.1571,
      "step": 20
    },
    {
      "epoch": 0.08,
      "learning_rate": 1e-05,
      "loss": 1.1534,
      "step": 21
    },
    {
      "epoch": 0.09,
      "learning_rate": 1e-05,
      "loss": 1.1374,
      "step": 22
    },
    {
      "epoch": 0.09,
      "learning_rate": 1e-05,
      "loss": 1.1084,
      "step": 23
    },
    {
      "epoch": 0.1,
      "learning_rate": 1e-05,
      "loss": 1.1097,
      "step": 24
    },
    {
      "epoch": 0.1,
      "learning_rate": 1e-05,
      "loss": 1.1086,
      "step": 25
    },
    {
      "epoch": 0.1,
      "learning_rate": 1e-05,
      "loss": 1.1136,
      "step": 26
    },
    {
      "epoch": 0.11,
      "learning_rate": 1e-05,
      "loss": 1.0891,
      "step": 27
    },
    {
      "epoch": 0.11,
      "learning_rate": 1e-05,
      "loss": 1.093,
      "step": 28
    },
    {
      "epoch": 0.11,
      "learning_rate": 1e-05,
      "loss": 1.0812,
      "step": 29
    },
    {
      "epoch": 0.12,
      "learning_rate": 1e-05,
      "loss": 1.0625,
      "step": 30
    },
    {
      "epoch": 0.12,
      "learning_rate": 1e-05,
      "loss": 1.0789,
      "step": 31
    },
    {
      "epoch": 0.13,
      "learning_rate": 1e-05,
      "loss": 1.0537,
      "step": 32
    },
    {
      "epoch": 0.13,
      "learning_rate": 1e-05,
      "loss": 1.0817,
      "step": 33
    },
    {
      "epoch": 0.13,
      "learning_rate": 1e-05,
      "loss": 1.0276,
      "step": 34
    },
    {
      "epoch": 0.14,
      "learning_rate": 1e-05,
      "loss": 1.019,
      "step": 35
    },
    {
      "epoch": 0.14,
      "learning_rate": 1e-05,
      "loss": 1.0213,
      "step": 36
    },
    {
      "epoch": 0.15,
      "learning_rate": 1e-05,
      "loss": 1.0408,
      "step": 37
    },
    {
      "epoch": 0.15,
      "learning_rate": 1e-05,
      "loss": 1.0321,
      "step": 38
    },
    {
      "epoch": 0.15,
      "learning_rate": 1e-05,
      "loss": 1.0372,
      "step": 39
    },
    {
      "epoch": 0.16,
      "learning_rate": 1e-05,
      "loss": 1.0137,
      "step": 40
    },
    {
      "epoch": 0.16,
      "learning_rate": 1e-05,
      "loss": 1.0127,
      "step": 41
    },
    {
      "epoch": 0.17,
      "learning_rate": 1e-05,
      "loss": 0.9953,
      "step": 42
    },
    {
      "epoch": 0.17,
      "learning_rate": 1e-05,
      "loss": 1.0146,
      "step": 43
    },
    {
      "epoch": 0.17,
      "learning_rate": 1e-05,
      "loss": 0.9715,
      "step": 44
    },
    {
      "epoch": 0.18,
      "learning_rate": 1e-05,
      "loss": 1.0089,
      "step": 45
    },
    {
      "epoch": 0.18,
      "learning_rate": 1e-05,
      "loss": 0.966,
      "step": 46
    },
    {
      "epoch": 0.19,
      "learning_rate": 1e-05,
      "loss": 0.9821,
      "step": 47
    },
    {
      "epoch": 0.19,
      "learning_rate": 1e-05,
      "loss": 0.9886,
      "step": 48
    },
    {
      "epoch": 0.19,
      "learning_rate": 1e-05,
      "loss": 0.9724,
      "step": 49
    },
    {
      "epoch": 0.2,
      "learning_rate": 1e-05,
      "loss": 0.9824,
      "step": 50
    },
    {
      "epoch": 0.2,
      "learning_rate": 1e-05,
      "loss": 0.9939,
      "step": 51
    },
    {
      "epoch": 0.21,
      "learning_rate": 1e-05,
      "loss": 0.9586,
      "step": 52
    },
    {
      "epoch": 0.21,
      "learning_rate": 1e-05,
      "loss": 0.9623,
      "step": 53
    },
    {
      "epoch": 0.21,
      "learning_rate": 1e-05,
      "loss": 0.9608,
      "step": 54
    },
    {
      "epoch": 0.22,
      "learning_rate": 1e-05,
      "loss": 0.9648,
      "step": 55
    },
    {
      "epoch": 0.22,
      "learning_rate": 1e-05,
      "loss": 0.9637,
      "step": 56
    },
    {
      "epoch": 0.23,
      "learning_rate": 1e-05,
      "loss": 0.9633,
      "step": 57
    },
    {
      "epoch": 0.23,
      "learning_rate": 1e-05,
      "loss": 0.9404,
      "step": 58
    },
    {
      "epoch": 0.23,
      "learning_rate": 1e-05,
      "loss": 0.9505,
      "step": 59
    },
    {
      "epoch": 0.24,
      "learning_rate": 1e-05,
      "loss": 0.952,
      "step": 60
    },
    {
      "epoch": 0.24,
      "learning_rate": 1e-05,
      "loss": 0.9445,
      "step": 61
    },
    {
      "epoch": 0.25,
      "learning_rate": 1e-05,
      "loss": 0.938,
      "step": 62
    },
    {
      "epoch": 0.25,
      "learning_rate": 1e-05,
      "loss": 0.9406,
      "step": 63
    },
    {
      "epoch": 0.25,
      "learning_rate": 1e-05,
      "loss": 0.9375,
      "step": 64
    },
    {
      "epoch": 0.26,
      "learning_rate": 1e-05,
      "loss": 0.9551,
      "step": 65
    },
    {
      "epoch": 0.26,
      "learning_rate": 1e-05,
      "loss": 0.9309,
      "step": 66
    },
    {
      "epoch": 0.27,
      "learning_rate": 1e-05,
      "loss": 0.9238,
      "step": 67
    },
    {
      "epoch": 0.27,
      "learning_rate": 1e-05,
      "loss": 0.9394,
      "step": 68
    },
    {
      "epoch": 0.27,
      "learning_rate": 1e-05,
      "loss": 0.9225,
      "step": 69
    },
    {
      "epoch": 0.28,
      "learning_rate": 1e-05,
      "loss": 0.9243,
      "step": 70
    },
    {
      "epoch": 0.28,
      "learning_rate": 1e-05,
      "loss": 0.9428,
      "step": 71
    },
    {
      "epoch": 0.29,
      "learning_rate": 1e-05,
      "loss": 0.9456,
      "step": 72
    },
    {
      "epoch": 0.29,
      "learning_rate": 1e-05,
      "loss": 0.9152,
      "step": 73
    },
    {
      "epoch": 0.29,
      "learning_rate": 1e-05,
      "loss": 0.9175,
      "step": 74
    },
    {
      "epoch": 0.3,
      "learning_rate": 1e-05,
      "loss": 0.9345,
      "step": 75
    },
    {
      "epoch": 0.3,
      "learning_rate": 1e-05,
      "loss": 0.9294,
      "step": 76
    },
    {
      "epoch": 0.31,
      "learning_rate": 1e-05,
      "loss": 0.9362,
      "step": 77
    },
    {
      "epoch": 0.31,
      "learning_rate": 1e-05,
      "loss": 0.926,
      "step": 78
    },
    {
      "epoch": 0.31,
      "learning_rate": 1e-05,
      "loss": 0.9268,
      "step": 79
    },
    {
      "epoch": 0.32,
      "learning_rate": 1e-05,
      "loss": 0.8947,
      "step": 80
    },
    {
      "epoch": 0.32,
      "learning_rate": 1e-05,
      "loss": 0.9137,
      "step": 81
    },
    {
      "epoch": 0.33,
      "learning_rate": 1e-05,
      "loss": 0.9229,
      "step": 82
    },
    {
      "epoch": 0.33,
      "learning_rate": 1e-05,
      "loss": 0.9264,
      "step": 83
    },
    {
      "epoch": 0.33,
      "learning_rate": 1e-05,
      "loss": 0.924,
      "step": 84
    },
    {
      "epoch": 0.34,
      "learning_rate": 1e-05,
      "loss": 0.9193,
      "step": 85
    },
    {
      "epoch": 0.34,
      "learning_rate": 1e-05,
      "loss": 0.8953,
      "step": 86
    },
    {
      "epoch": 0.34,
      "learning_rate": 1e-05,
      "loss": 0.9146,
      "step": 87
    },
    {
      "epoch": 0.35,
      "learning_rate": 1e-05,
      "loss": 0.8716,
      "step": 88
    },
    {
      "epoch": 0.35,
      "learning_rate": 1e-05,
      "loss": 0.9132,
      "step": 89
    },
    {
      "epoch": 0.36,
      "learning_rate": 1e-05,
      "loss": 0.9098,
      "step": 90
    },
    {
      "epoch": 0.36,
      "learning_rate": 1e-05,
      "loss": 0.9142,
      "step": 91
    },
    {
      "epoch": 0.36,
      "learning_rate": 1e-05,
      "loss": 0.8964,
      "step": 92
    },
    {
      "epoch": 0.37,
      "learning_rate": 1e-05,
      "loss": 0.9001,
      "step": 93
    },
    {
      "epoch": 0.37,
      "learning_rate": 1e-05,
      "loss": 0.9355,
      "step": 94
    },
    {
      "epoch": 0.38,
      "learning_rate": 1e-05,
      "loss": 0.9194,
      "step": 95
    },
    {
      "epoch": 0.38,
      "learning_rate": 1e-05,
      "loss": 0.8819,
      "step": 96
    },
    {
      "epoch": 0.38,
      "learning_rate": 1e-05,
      "loss": 0.8973,
      "step": 97
    },
    {
      "epoch": 0.39,
      "learning_rate": 1e-05,
      "loss": 0.8867,
      "step": 98
    },
    {
      "epoch": 0.39,
      "learning_rate": 1e-05,
      "loss": 0.9023,
      "step": 99
    },
    {
      "epoch": 0.4,
      "learning_rate": 1e-05,
      "loss": 0.8993,
      "step": 100
    },
    {
      "epoch": 0.4,
      "learning_rate": 1e-05,
      "loss": 0.8881,
      "step": 101
    },
    {
      "epoch": 0.4,
      "learning_rate": 1e-05,
      "loss": 0.8819,
      "step": 102
    },
    {
      "epoch": 0.41,
      "learning_rate": 1e-05,
      "loss": 0.8875,
      "step": 103
    },
    {
      "epoch": 0.41,
      "learning_rate": 1e-05,
      "loss": 0.8871,
      "step": 104
    },
    {
      "epoch": 0.42,
      "learning_rate": 1e-05,
      "loss": 0.8898,
      "step": 105
    },
    {
      "epoch": 0.42,
      "learning_rate": 1e-05,
      "loss": 0.8707,
      "step": 106
    },
    {
      "epoch": 0.42,
      "learning_rate": 1e-05,
      "loss": 0.8959,
      "step": 107
    },
    {
      "epoch": 0.43,
      "learning_rate": 1e-05,
      "loss": 0.8812,
      "step": 108
    },
    {
      "epoch": 0.43,
      "learning_rate": 1e-05,
      "loss": 0.8955,
      "step": 109
    },
    {
      "epoch": 0.44,
      "learning_rate": 1e-05,
      "loss": 0.8712,
      "step": 110
    },
    {
      "epoch": 0.44,
      "learning_rate": 1e-05,
      "loss": 0.9046,
      "step": 111
    },
    {
      "epoch": 0.44,
      "learning_rate": 1e-05,
      "loss": 0.8832,
      "step": 112
    },
    {
      "epoch": 0.45,
      "learning_rate": 1e-05,
      "loss": 0.9039,
      "step": 113
    },
    {
      "epoch": 0.45,
      "learning_rate": 1e-05,
      "loss": 0.8783,
      "step": 114
    },
    {
      "epoch": 0.46,
      "learning_rate": 1e-05,
      "loss": 0.8819,
      "step": 115
    },
    {
      "epoch": 0.46,
      "learning_rate": 1e-05,
      "loss": 0.873,
      "step": 116
    },
    {
      "epoch": 0.46,
      "learning_rate": 1e-05,
      "loss": 0.8628,
      "step": 117
    },
    {
      "epoch": 0.47,
      "learning_rate": 1e-05,
      "loss": 0.8649,
      "step": 118
    },
    {
      "epoch": 0.47,
      "learning_rate": 1e-05,
      "loss": 0.89,
      "step": 119
    },
    {
      "epoch": 0.48,
      "learning_rate": 1e-05,
      "loss": 0.8864,
      "step": 120
    },
    {
      "epoch": 0.48,
      "learning_rate": 1e-05,
      "loss": 0.8653,
      "step": 121
    },
    {
      "epoch": 0.48,
      "learning_rate": 1e-05,
      "loss": 0.865,
      "step": 122
    },
    {
      "epoch": 0.49,
      "learning_rate": 1e-05,
      "loss": 0.8448,
      "step": 123
    },
    {
      "epoch": 0.49,
      "learning_rate": 1e-05,
      "loss": 0.8363,
      "step": 124
    },
    {
      "epoch": 0.5,
      "learning_rate": 1e-05,
      "loss": 0.8586,
      "step": 125
    },
    {
      "epoch": 0.5,
      "learning_rate": 1e-05,
      "loss": 0.8603,
      "step": 126
    },
    {
      "epoch": 0.5,
      "learning_rate": 1e-05,
      "loss": 0.8784,
      "step": 127
    },
    {
      "epoch": 0.51,
      "learning_rate": 1e-05,
      "loss": 0.8822,
      "step": 128
    },
    {
      "epoch": 0.51,
      "learning_rate": 1e-05,
      "loss": 0.8592,
      "step": 129
    },
    {
      "epoch": 0.52,
      "learning_rate": 1e-05,
      "loss": 0.8629,
      "step": 130
    },
    {
      "epoch": 0.52,
      "learning_rate": 1e-05,
      "loss": 0.8691,
      "step": 131
    },
    {
      "epoch": 0.52,
      "learning_rate": 1e-05,
      "loss": 0.8684,
      "step": 132
    },
    {
      "epoch": 0.53,
      "learning_rate": 1e-05,
      "loss": 0.8767,
      "step": 133
    },
    {
      "epoch": 0.53,
      "learning_rate": 1e-05,
      "loss": 0.8809,
      "step": 134
    },
    {
      "epoch": 0.54,
      "learning_rate": 1e-05,
      "loss": 0.8504,
      "step": 135
    },
    {
      "epoch": 0.54,
      "learning_rate": 1e-05,
      "loss": 0.8705,
      "step": 136
    },
    {
      "epoch": 0.54,
      "learning_rate": 1e-05,
      "loss": 0.8721,
      "step": 137
    },
    {
      "epoch": 0.55,
      "learning_rate": 1e-05,
      "loss": 0.8588,
      "step": 138
    },
    {
      "epoch": 0.55,
      "learning_rate": 1e-05,
      "loss": 0.8842,
      "step": 139
    },
    {
      "epoch": 0.56,
      "learning_rate": 1e-05,
      "loss": 0.8553,
      "step": 140
    },
    {
      "epoch": 0.56,
      "learning_rate": 1e-05,
      "loss": 0.8546,
      "step": 141
    },
    {
      "epoch": 0.56,
      "learning_rate": 1e-05,
      "loss": 0.8588,
      "step": 142
    },
    {
      "epoch": 0.57,
      "learning_rate": 1e-05,
      "loss": 0.8672,
      "step": 143
    },
    {
      "epoch": 0.57,
      "learning_rate": 1e-05,
      "loss": 0.8745,
      "step": 144
    },
    {
      "epoch": 0.57,
      "learning_rate": 1e-05,
      "loss": 0.8842,
      "step": 145
    },
    {
      "epoch": 0.58,
      "learning_rate": 1e-05,
      "loss": 0.8594,
      "step": 146
    },
    {
      "epoch": 0.58,
      "learning_rate": 1e-05,
      "loss": 0.8463,
      "step": 147
    },
    {
      "epoch": 0.59,
      "learning_rate": 1e-05,
      "loss": 0.8399,
      "step": 148
    },
    {
      "epoch": 0.59,
      "learning_rate": 1e-05,
      "loss": 0.8521,
      "step": 149
    },
    {
      "epoch": 0.59,
      "learning_rate": 1e-05,
      "loss": 0.8492,
      "step": 150
    },
    {
      "epoch": 0.6,
      "learning_rate": 1e-05,
      "loss": 0.8535,
      "step": 151
    },
    {
      "epoch": 0.6,
      "learning_rate": 1e-05,
      "loss": 0.8552,
      "step": 152
    },
    {
      "epoch": 0.61,
      "learning_rate": 1e-05,
      "loss": 0.8602,
      "step": 153
    },
    {
      "epoch": 0.61,
      "learning_rate": 1e-05,
      "loss": 0.8559,
      "step": 154
    },
    {
      "epoch": 0.61,
      "learning_rate": 1e-05,
      "loss": 0.8498,
      "step": 155
    },
    {
      "epoch": 0.62,
      "learning_rate": 1e-05,
      "loss": 0.832,
      "step": 156
    },
    {
      "epoch": 0.62,
      "learning_rate": 1e-05,
      "loss": 0.8518,
      "step": 157
    },
    {
      "epoch": 0.63,
      "learning_rate": 1e-05,
      "loss": 0.8296,
      "step": 158
    },
    {
      "epoch": 0.63,
      "learning_rate": 1e-05,
      "loss": 0.8346,
      "step": 159
    },
    {
      "epoch": 0.63,
      "learning_rate": 1e-05,
      "loss": 0.8491,
      "step": 160
    },
    {
      "epoch": 0.64,
      "learning_rate": 1e-05,
      "loss": 0.8432,
      "step": 161
    },
    {
      "epoch": 0.64,
      "learning_rate": 1e-05,
      "loss": 0.842,
      "step": 162
    },
    {
      "epoch": 0.65,
      "learning_rate": 1e-05,
      "loss": 0.8618,
      "step": 163
    },
    {
      "epoch": 0.65,
      "learning_rate": 1e-05,
      "loss": 0.8284,
      "step": 164
    },
    {
      "epoch": 0.65,
      "learning_rate": 1e-05,
      "loss": 0.8648,
      "step": 165
    },
    {
      "epoch": 0.66,
      "learning_rate": 1e-05,
      "loss": 0.8601,
      "step": 166
    },
    {
      "epoch": 0.66,
      "learning_rate": 1e-05,
      "loss": 0.8533,
      "step": 167
    },
    {
      "epoch": 0.67,
      "learning_rate": 1e-05,
      "loss": 0.8414,
      "step": 168
    },
    {
      "epoch": 0.67,
      "learning_rate": 1e-05,
      "loss": 0.8386,
      "step": 169
    },
    {
      "epoch": 0.67,
      "learning_rate": 1e-05,
      "loss": 0.8454,
      "step": 170
    },
    {
      "epoch": 0.68,
      "learning_rate": 1e-05,
      "loss": 0.8384,
      "step": 171
    },
    {
      "epoch": 0.68,
      "learning_rate": 1e-05,
      "loss": 0.8452,
      "step": 172
    },
    {
      "epoch": 0.69,
      "learning_rate": 1e-05,
      "loss": 0.8445,
      "step": 173
    },
    {
      "epoch": 0.69,
      "learning_rate": 1e-05,
      "loss": 0.8499,
      "step": 174
    },
    {
      "epoch": 0.69,
      "learning_rate": 1e-05,
      "loss": 0.8417,
      "step": 175
    },
    {
      "epoch": 0.7,
      "learning_rate": 1e-05,
      "loss": 0.8419,
      "step": 176
    },
    {
      "epoch": 0.7,
      "learning_rate": 1e-05,
      "loss": 0.834,
      "step": 177
    },
    {
      "epoch": 0.71,
      "learning_rate": 1e-05,
      "loss": 0.8281,
      "step": 178
    },
    {
      "epoch": 0.71,
      "learning_rate": 1e-05,
      "loss": 0.8398,
      "step": 179
    },
    {
      "epoch": 0.71,
      "learning_rate": 1e-05,
      "loss": 0.87,
      "step": 180
    },
    {
      "epoch": 0.72,
      "learning_rate": 1e-05,
      "loss": 0.8256,
      "step": 181
    },
    {
      "epoch": 0.72,
      "learning_rate": 1e-05,
      "loss": 0.8087,
      "step": 182
    },
    {
      "epoch": 0.73,
      "learning_rate": 1e-05,
      "loss": 0.8268,
      "step": 183
    },
    {
      "epoch": 0.73,
      "learning_rate": 1e-05,
      "loss": 0.8269,
      "step": 184
    },
    {
      "epoch": 0.73,
      "learning_rate": 1e-05,
      "loss": 0.8021,
      "step": 185
    },
    {
      "epoch": 0.74,
      "learning_rate": 1e-05,
      "loss": 0.8157,
      "step": 186
    },
    {
      "epoch": 0.74,
      "learning_rate": 1e-05,
      "loss": 0.8319,
      "step": 187
    },
    {
      "epoch": 0.75,
      "learning_rate": 1e-05,
      "loss": 0.7985,
      "step": 188
    },
    {
      "epoch": 0.75,
      "learning_rate": 1e-05,
      "loss": 0.8481,
      "step": 189
    },
    {
      "epoch": 0.75,
      "learning_rate": 1e-05,
      "loss": 0.832,
      "step": 190
    },
    {
      "epoch": 0.76,
      "learning_rate": 1e-05,
      "loss": 0.8245,
      "step": 191
    },
    {
      "epoch": 0.76,
      "learning_rate": 1e-05,
      "loss": 0.8116,
      "step": 192
    },
    {
      "epoch": 0.77,
      "learning_rate": 1e-05,
      "loss": 0.8171,
      "step": 193
    },
    {
      "epoch": 0.77,
      "learning_rate": 1e-05,
      "loss": 0.8091,
      "step": 194
    },
    {
      "epoch": 0.77,
      "learning_rate": 1e-05,
      "loss": 0.8032,
      "step": 195
    },
    {
      "epoch": 0.78,
      "learning_rate": 1e-05,
      "loss": 0.8082,
      "step": 196
    },
    {
      "epoch": 0.78,
      "learning_rate": 1e-05,
      "loss": 0.8279,
      "step": 197
    },
    {
      "epoch": 0.79,
      "learning_rate": 1e-05,
      "loss": 0.8574,
      "step": 198
    },
    {
      "epoch": 0.79,
      "learning_rate": 1e-05,
      "loss": 0.8457,
      "step": 199
    },
    {
      "epoch": 0.79,
      "learning_rate": 1e-05,
      "loss": 0.8208,
      "step": 200
    },
    {
      "epoch": 0.8,
      "learning_rate": 1e-05,
      "loss": 0.8238,
      "step": 201
    },
    {
      "epoch": 0.8,
      "learning_rate": 1e-05,
      "loss": 0.8417,
      "step": 202
    },
    {
      "epoch": 0.8,
      "learning_rate": 1e-05,
      "loss": 0.809,
      "step": 203
    },
    {
      "epoch": 0.81,
      "learning_rate": 1e-05,
      "loss": 0.825,
      "step": 204
    },
    {
      "epoch": 0.81,
      "learning_rate": 1e-05,
      "loss": 0.8171,
      "step": 205
    },
    {
      "epoch": 0.82,
      "learning_rate": 1e-05,
      "loss": 0.8174,
      "step": 206
    },
    {
      "epoch": 0.82,
      "learning_rate": 1e-05,
      "loss": 0.8167,
      "step": 207
    },
    {
      "epoch": 0.82,
      "learning_rate": 1e-05,
      "loss": 0.8316,
      "step": 208
    },
    {
      "epoch": 0.83,
      "learning_rate": 1e-05,
      "loss": 0.8039,
      "step": 209
    },
    {
      "epoch": 0.83,
      "learning_rate": 1e-05,
      "loss": 0.8137,
      "step": 210
    },
    {
      "epoch": 0.84,
      "learning_rate": 1e-05,
      "loss": 0.8183,
      "step": 211
    },
    {
      "epoch": 0.84,
      "learning_rate": 1e-05,
      "loss": 0.8278,
      "step": 212
    },
    {
      "epoch": 0.84,
      "learning_rate": 1e-05,
      "loss": 0.8055,
      "step": 213
    },
    {
      "epoch": 0.85,
      "learning_rate": 1e-05,
      "loss": 0.7998,
      "step": 214
    },
    {
      "epoch": 0.85,
      "learning_rate": 1e-05,
      "loss": 0.8213,
      "step": 215
    },
    {
      "epoch": 0.86,
      "learning_rate": 1e-05,
      "loss": 0.796,
      "step": 216
    },
    {
      "epoch": 0.86,
      "learning_rate": 1e-05,
      "loss": 0.8069,
      "step": 217
    },
    {
      "epoch": 0.86,
      "learning_rate": 1e-05,
      "loss": 0.8021,
      "step": 218
    },
    {
      "epoch": 0.87,
      "learning_rate": 1e-05,
      "loss": 0.8076,
      "step": 219
    },
    {
      "epoch": 0.87,
      "learning_rate": 1e-05,
      "loss": 0.8225,
      "step": 220
    },
    {
      "epoch": 0.88,
      "learning_rate": 1e-05,
      "loss": 0.8025,
      "step": 221
    },
    {
      "epoch": 0.88,
      "learning_rate": 1e-05,
      "loss": 0.8132,
      "step": 222
    },
    {
      "epoch": 0.88,
      "learning_rate": 1e-05,
      "loss": 0.8062,
      "step": 223
    },
    {
      "epoch": 0.89,
      "learning_rate": 1e-05,
      "loss": 0.8243,
      "step": 224
    },
    {
      "epoch": 0.89,
      "learning_rate": 1e-05,
      "loss": 0.8204,
      "step": 225
    },
    {
      "epoch": 0.9,
      "learning_rate": 1e-05,
      "loss": 0.8211,
      "step": 226
    },
    {
      "epoch": 0.9,
      "learning_rate": 1e-05,
      "loss": 0.8167,
      "step": 227
    },
    {
      "epoch": 0.9,
      "learning_rate": 1e-05,
      "loss": 0.7995,
      "step": 228
    },
    {
      "epoch": 0.91,
      "learning_rate": 1e-05,
      "loss": 0.7867,
      "step": 229
    },
    {
      "epoch": 0.91,
      "learning_rate": 1e-05,
      "loss": 0.7931,
      "step": 230
    },
    {
      "epoch": 0.92,
      "learning_rate": 1e-05,
      "loss": 0.7975,
      "step": 231
    },
    {
      "epoch": 0.92,
      "learning_rate": 1e-05,
      "loss": 0.8271,
      "step": 232
    },
    {
      "epoch": 0.92,
      "learning_rate": 1e-05,
      "loss": 0.7776,
      "step": 233
    },
    {
      "epoch": 0.93,
      "learning_rate": 1e-05,
      "loss": 0.7944,
      "step": 234
    },
    {
      "epoch": 0.93,
      "learning_rate": 1e-05,
      "loss": 0.8222,
      "step": 235
    },
    {
      "epoch": 0.94,
      "learning_rate": 1e-05,
      "loss": 0.797,
      "step": 236
    },
    {
      "epoch": 0.94,
      "learning_rate": 1e-05,
      "loss": 0.8176,
      "step": 237
    },
    {
      "epoch": 0.94,
      "learning_rate": 1e-05,
      "loss": 0.7953,
      "step": 238
    },
    {
      "epoch": 0.95,
      "learning_rate": 1e-05,
      "loss": 0.8076,
      "step": 239
    },
    {
      "epoch": 0.95,
      "learning_rate": 1e-05,
      "loss": 0.7911,
      "step": 240
    },
    {
      "epoch": 0.96,
      "learning_rate": 1e-05,
      "loss": 0.8091,
      "step": 241
    },
    {
      "epoch": 0.96,
      "learning_rate": 1e-05,
      "loss": 0.7754,
      "step": 242
    },
    {
      "epoch": 0.96,
      "learning_rate": 1e-05,
      "loss": 0.7923,
      "step": 243
    },
    {
      "epoch": 0.97,
      "learning_rate": 1e-05,
      "loss": 0.8013,
      "step": 244
    },
    {
      "epoch": 0.97,
      "learning_rate": 1e-05,
      "loss": 0.7873,
      "step": 245
    },
    {
      "epoch": 0.98,
      "learning_rate": 1e-05,
      "loss": 0.7904,
      "step": 246
    },
    {
      "epoch": 0.98,
      "learning_rate": 1e-05,
      "loss": 0.8298,
      "step": 247
    },
    {
      "epoch": 0.98,
      "learning_rate": 1e-05,
      "loss": 0.8002,
      "step": 248
    },
    {
      "epoch": 0.99,
      "learning_rate": 1e-05,
      "loss": 0.7924,
      "step": 249
    },
    {
      "epoch": 0.99,
      "learning_rate": 1e-05,
      "loss": 0.7824,
      "step": 250
    },
    {
      "epoch": 1.0,
      "learning_rate": 1e-05,
      "loss": 0.7918,
      "step": 251
    },
    {
      "epoch": 1.0,
      "learning_rate": 1e-05,
      "loss": 0.8046,
      "step": 252
    },
    {
      "epoch": 1.0,
      "learning_rate": 1e-05,
      "loss": 0.7592,
      "step": 253
    },
    {
      "epoch": 1.01,
      "learning_rate": 1e-05,
      "loss": 0.7253,
      "step": 254
    },
    {
      "epoch": 1.01,
      "learning_rate": 1e-05,
      "loss": 0.7204,
      "step": 255
    },
    {
      "epoch": 1.02,
      "learning_rate": 1e-05,
      "loss": 0.7307,
      "step": 256
    },
    {
      "epoch": 1.02,
      "learning_rate": 1e-05,
      "loss": 0.7084,
      "step": 257
    },
    {
      "epoch": 1.02,
      "learning_rate": 1e-05,
      "loss": 0.7474,
      "step": 258
    },
    {
      "epoch": 1.03,
      "learning_rate": 1e-05,
      "loss": 0.7248,
      "step": 259
    },
    {
      "epoch": 1.03,
      "learning_rate": 1e-05,
      "loss": 0.7243,
      "step": 260
    },
    {
      "epoch": 1.03,
      "learning_rate": 1e-05,
      "loss": 0.7369,
      "step": 261
    },
    {
      "epoch": 1.04,
      "learning_rate": 1e-05,
      "loss": 0.7254,
      "step": 262
    },
    {
      "epoch": 1.04,
      "learning_rate": 1e-05,
      "loss": 0.7264,
      "step": 263
    },
    {
      "epoch": 1.05,
      "learning_rate": 1e-05,
      "loss": 0.6911,
      "step": 264
    },
    {
      "epoch": 1.05,
      "learning_rate": 1e-05,
      "loss": 0.7191,
      "step": 265
    },
    {
      "epoch": 1.05,
      "learning_rate": 1e-05,
      "loss": 0.717,
      "step": 266
    },
    {
      "epoch": 1.06,
      "learning_rate": 1e-05,
      "loss": 0.737,
      "step": 267
    },
    {
      "epoch": 1.06,
      "learning_rate": 1e-05,
      "loss": 0.7278,
      "step": 268
    },
    {
      "epoch": 1.07,
      "learning_rate": 1e-05,
      "loss": 0.7381,
      "step": 269
    },
    {
      "epoch": 1.07,
      "learning_rate": 1e-05,
      "loss": 0.709,
      "step": 270
    },
    {
      "epoch": 1.07,
      "learning_rate": 1e-05,
      "loss": 0.7021,
      "step": 271
    },
    {
      "epoch": 1.08,
      "learning_rate": 1e-05,
      "loss": 0.7472,
      "step": 272
    },
    {
      "epoch": 1.08,
      "learning_rate": 1e-05,
      "loss": 0.7367,
      "step": 273
    },
    {
      "epoch": 1.09,
      "learning_rate": 1e-05,
      "loss": 0.7065,
      "step": 274
    },
    {
      "epoch": 1.09,
      "learning_rate": 1e-05,
      "loss": 0.7272,
      "step": 275
    },
    {
      "epoch": 1.09,
      "learning_rate": 1e-05,
      "loss": 0.7133,
      "step": 276
    },
    {
      "epoch": 1.1,
      "learning_rate": 1e-05,
      "loss": 0.7388,
      "step": 277
    },
    {
      "epoch": 1.1,
      "learning_rate": 1e-05,
      "loss": 0.6945,
      "step": 278
    },
    {
      "epoch": 1.11,
      "learning_rate": 1e-05,
      "loss": 0.718,
      "step": 279
    },
    {
      "epoch": 1.11,
      "learning_rate": 1e-05,
      "loss": 0.7049,
      "step": 280
    },
    {
      "epoch": 1.11,
      "learning_rate": 1e-05,
      "loss": 0.7176,
      "step": 281
    },
    {
      "epoch": 1.12,
      "learning_rate": 1e-05,
      "loss": 0.722,
      "step": 282
    },
    {
      "epoch": 1.12,
      "learning_rate": 1e-05,
      "loss": 0.7099,
      "step": 283
    },
    {
      "epoch": 1.13,
      "learning_rate": 1e-05,
      "loss": 0.6972,
      "step": 284
    },
    {
      "epoch": 1.13,
      "learning_rate": 1e-05,
      "loss": 0.7266,
      "step": 285
    },
    {
      "epoch": 1.13,
      "learning_rate": 1e-05,
      "loss": 0.7021,
      "step": 286
    },
    {
      "epoch": 1.14,
      "learning_rate": 1e-05,
      "loss": 0.7085,
      "step": 287
    },
    {
      "epoch": 1.14,
      "learning_rate": 1e-05,
      "loss": 0.7117,
      "step": 288
    },
    {
      "epoch": 1.15,
      "learning_rate": 1e-05,
      "loss": 0.7316,
      "step": 289
    },
    {
      "epoch": 1.15,
      "learning_rate": 1e-05,
      "loss": 0.705,
      "step": 290
    },
    {
      "epoch": 1.15,
      "learning_rate": 1e-05,
      "loss": 0.68,
      "step": 291
    },
    {
      "epoch": 1.16,
      "learning_rate": 1e-05,
      "loss": 0.711,
      "step": 292
    },
    {
      "epoch": 1.16,
      "learning_rate": 1e-05,
      "loss": 0.7088,
      "step": 293
    },
    {
      "epoch": 1.17,
      "learning_rate": 1e-05,
      "loss": 0.683,
      "step": 294
    },
    {
      "epoch": 1.17,
      "learning_rate": 1e-05,
      "loss": 0.6971,
      "step": 295
    },
    {
      "epoch": 1.17,
      "learning_rate": 1e-05,
      "loss": 0.6915,
      "step": 296
    },
    {
      "epoch": 1.18,
      "learning_rate": 1e-05,
      "loss": 0.7053,
      "step": 297
    },
    {
      "epoch": 1.18,
      "learning_rate": 1e-05,
      "loss": 0.7176,
      "step": 298
    },
    {
      "epoch": 1.19,
      "learning_rate": 1e-05,
      "loss": 0.7129,
      "step": 299
    },
    {
      "epoch": 1.19,
      "learning_rate": 1e-05,
      "loss": 0.7166,
      "step": 300
    },
    {
      "epoch": 1.19,
      "learning_rate": 1e-05,
      "loss": 0.7275,
      "step": 301
    },
    {
      "epoch": 1.2,
      "learning_rate": 1e-05,
      "loss": 0.6941,
      "step": 302
    },
    {
      "epoch": 1.2,
      "learning_rate": 1e-05,
      "loss": 0.7153,
      "step": 303
    },
    {
      "epoch": 1.21,
      "learning_rate": 1e-05,
      "loss": 0.7449,
      "step": 304
    },
    {
      "epoch": 1.21,
      "learning_rate": 1e-05,
      "loss": 0.7393,
      "step": 305
    },
    {
      "epoch": 1.21,
      "learning_rate": 1e-05,
      "loss": 0.7146,
      "step": 306
    },
    {
      "epoch": 1.22,
      "learning_rate": 1e-05,
      "loss": 0.6976,
      "step": 307
    },
    {
      "epoch": 1.22,
      "learning_rate": 1e-05,
      "loss": 0.7072,
      "step": 308
    },
    {
      "epoch": 1.23,
      "learning_rate": 1e-05,
      "loss": 0.677,
      "step": 309
    },
    {
      "epoch": 1.23,
      "learning_rate": 1e-05,
      "loss": 0.7155,
      "step": 310
    },
    {
      "epoch": 1.23,
      "learning_rate": 1e-05,
      "loss": 0.7057,
      "step": 311
    },
    {
      "epoch": 1.24,
      "learning_rate": 1e-05,
      "loss": 0.7008,
      "step": 312
    },
    {
      "epoch": 1.24,
      "learning_rate": 1e-05,
      "loss": 0.7123,
      "step": 313
    },
    {
      "epoch": 1.25,
      "learning_rate": 1e-05,
      "loss": 0.6998,
      "step": 314
    },
    {
      "epoch": 1.25,
      "learning_rate": 1e-05,
      "loss": 0.7242,
      "step": 315
    },
    {
      "epoch": 1.25,
      "learning_rate": 1e-05,
      "loss": 0.7009,
      "step": 316
    },
    {
      "epoch": 1.26,
      "learning_rate": 1e-05,
      "loss": 0.7019,
      "step": 317
    },
    {
      "epoch": 1.26,
      "learning_rate": 1e-05,
      "loss": 0.7173,
      "step": 318
    },
    {
      "epoch": 1.26,
      "learning_rate": 1e-05,
      "loss": 0.6784,
      "step": 319
    },
    {
      "epoch": 1.27,
      "learning_rate": 1e-05,
      "loss": 0.6833,
      "step": 320
    },
    {
      "epoch": 1.27,
      "learning_rate": 1e-05,
      "loss": 0.6952,
      "step": 321
    },
    {
      "epoch": 1.28,
      "learning_rate": 1e-05,
      "loss": 0.6893,
      "step": 322
    },
    {
      "epoch": 1.28,
      "learning_rate": 1e-05,
      "loss": 0.6864,
      "step": 323
    },
    {
      "epoch": 1.28,
      "learning_rate": 1e-05,
      "loss": 0.691,
      "step": 324
    },
    {
      "epoch": 1.29,
      "learning_rate": 1e-05,
      "loss": 0.701,
      "step": 325
    },
    {
      "epoch": 1.29,
      "learning_rate": 1e-05,
      "loss": 0.7076,
      "step": 326
    },
    {
      "epoch": 1.3,
      "learning_rate": 1e-05,
      "loss": 0.6974,
      "step": 327
    },
    {
      "epoch": 1.3,
      "learning_rate": 1e-05,
      "loss": 0.7064,
      "step": 328
    },
    {
      "epoch": 1.3,
      "learning_rate": 1e-05,
      "loss": 0.6979,
      "step": 329
    },
    {
      "epoch": 1.31,
      "learning_rate": 1e-05,
      "loss": 0.6983,
      "step": 330
    },
    {
      "epoch": 1.31,
      "learning_rate": 1e-05,
      "loss": 0.7082,
      "step": 331
    },
    {
      "epoch": 1.32,
      "learning_rate": 1e-05,
      "loss": 0.7141,
      "step": 332
    },
    {
      "epoch": 1.32,
      "learning_rate": 1e-05,
      "loss": 0.7016,
      "step": 333
    },
    {
      "epoch": 1.32,
      "learning_rate": 1e-05,
      "loss": 0.7184,
      "step": 334
    },
    {
      "epoch": 1.33,
      "learning_rate": 1e-05,
      "loss": 0.731,
      "step": 335
    },
    {
      "epoch": 1.33,
      "learning_rate": 1e-05,
      "loss": 0.6935,
      "step": 336
    },
    {
      "epoch": 1.34,
      "learning_rate": 1e-05,
      "loss": 0.6983,
      "step": 337
    },
    {
      "epoch": 1.34,
      "learning_rate": 1e-05,
      "loss": 0.6945,
      "step": 338
    },
    {
      "epoch": 1.34,
      "learning_rate": 1e-05,
      "loss": 0.6929,
      "step": 339
    },
    {
      "epoch": 1.35,
      "learning_rate": 1e-05,
      "loss": 0.7104,
      "step": 340
    },
    {
      "epoch": 1.35,
      "learning_rate": 1e-05,
      "loss": 0.7037,
      "step": 341
    },
    {
      "epoch": 1.36,
      "learning_rate": 1e-05,
      "loss": 0.7152,
      "step": 342
    },
    {
      "epoch": 1.36,
      "learning_rate": 1e-05,
      "loss": 0.6899,
      "step": 343
    },
    {
      "epoch": 1.36,
      "learning_rate": 1e-05,
      "loss": 0.6838,
      "step": 344
    },
    {
      "epoch": 1.37,
      "learning_rate": 1e-05,
      "loss": 0.6725,
      "step": 345
    },
    {
      "epoch": 1.37,
      "learning_rate": 1e-05,
      "loss": 0.7175,
      "step": 346
    },
    {
      "epoch": 1.38,
      "learning_rate": 1e-05,
      "loss": 0.6856,
      "step": 347
    },
    {
      "epoch": 1.38,
      "learning_rate": 1e-05,
      "loss": 0.6924,
      "step": 348
    },
    {
      "epoch": 1.38,
      "learning_rate": 1e-05,
      "loss": 0.6914,
      "step": 349
    },
    {
      "epoch": 1.39,
      "learning_rate": 1e-05,
      "loss": 0.6825,
      "step": 350
    },
    {
      "epoch": 1.39,
      "learning_rate": 1e-05,
      "loss": 0.6952,
      "step": 351
    },
    {
      "epoch": 1.4,
      "learning_rate": 1e-05,
      "loss": 0.7079,
      "step": 352
    },
    {
      "epoch": 1.4,
      "learning_rate": 1e-05,
      "loss": 0.6956,
      "step": 353
    },
    {
      "epoch": 1.4,
      "learning_rate": 1e-05,
      "loss": 0.7019,
      "step": 354
    },
    {
      "epoch": 1.41,
      "learning_rate": 1e-05,
      "loss": 0.6973,
      "step": 355
    },
    {
      "epoch": 1.41,
      "learning_rate": 1e-05,
      "loss": 0.6784,
      "step": 356
    },
    {
      "epoch": 1.42,
      "learning_rate": 1e-05,
      "loss": 0.6913,
      "step": 357
    },
    {
      "epoch": 1.42,
      "learning_rate": 1e-05,
      "loss": 0.6872,
      "step": 358
    },
    {
      "epoch": 1.42,
      "learning_rate": 1e-05,
      "loss": 0.7088,
      "step": 359
    },
    {
      "epoch": 1.43,
      "learning_rate": 1e-05,
      "loss": 0.6955,
      "step": 360
    },
    {
      "epoch": 1.43,
      "learning_rate": 1e-05,
      "loss": 0.6927,
      "step": 361
    },
    {
      "epoch": 1.44,
      "learning_rate": 1e-05,
      "loss": 0.6839,
      "step": 362
    },
    {
      "epoch": 1.44,
      "learning_rate": 1e-05,
      "loss": 0.6717,
      "step": 363
    },
    {
      "epoch": 1.44,
      "learning_rate": 1e-05,
      "loss": 0.7001,
      "step": 364
    },
    {
      "epoch": 1.45,
      "learning_rate": 1e-05,
      "loss": 0.6687,
      "step": 365
    },
    {
      "epoch": 1.45,
      "learning_rate": 1e-05,
      "loss": 0.7016,
      "step": 366
    },
    {
      "epoch": 1.46,
      "learning_rate": 1e-05,
      "loss": 0.6807,
      "step": 367
    },
    {
      "epoch": 1.46,
      "learning_rate": 1e-05,
      "loss": 0.7073,
      "step": 368
    },
    {
      "epoch": 1.46,
      "learning_rate": 1e-05,
      "loss": 0.6704,
      "step": 369
    },
    {
      "epoch": 1.47,
      "learning_rate": 1e-05,
      "loss": 0.6812,
      "step": 370
    },
    {
      "epoch": 1.47,
      "learning_rate": 1e-05,
      "loss": 0.6579,
      "step": 371
    },
    {
      "epoch": 1.48,
      "learning_rate": 1e-05,
      "loss": 0.6825,
      "step": 372
    },
    {
      "epoch": 1.48,
      "learning_rate": 1e-05,
      "loss": 0.7014,
      "step": 373
    },
    {
      "epoch": 1.48,
      "learning_rate": 1e-05,
      "loss": 0.6789,
      "step": 374
    },
    {
      "epoch": 1.49,
      "learning_rate": 1e-05,
      "loss": 0.6753,
      "step": 375
    },
    {
      "epoch": 1.49,
      "learning_rate": 1e-05,
      "loss": 0.678,
      "step": 376
    },
    {
      "epoch": 1.49,
      "learning_rate": 1e-05,
      "loss": 0.6874,
      "step": 377
    },
    {
      "epoch": 1.5,
      "learning_rate": 1e-05,
      "loss": 0.6871,
      "step": 378
    },
    {
      "epoch": 1.5,
      "learning_rate": 1e-05,
      "loss": 0.6877,
      "step": 379
    },
    {
      "epoch": 1.51,
      "learning_rate": 1e-05,
      "loss": 0.7023,
      "step": 380
    },
    {
      "epoch": 1.51,
      "learning_rate": 1e-05,
      "loss": 0.7043,
      "step": 381
    },
    {
      "epoch": 1.51,
      "learning_rate": 1e-05,
      "loss": 0.6916,
      "step": 382
    },
    {
      "epoch": 1.52,
      "learning_rate": 1e-05,
      "loss": 0.6834,
      "step": 383
    },
    {
      "epoch": 1.52,
      "learning_rate": 1e-05,
      "loss": 0.6896,
      "step": 384
    },
    {
      "epoch": 1.53,
      "learning_rate": 1e-05,
      "loss": 0.6738,
      "step": 385
    },
    {
      "epoch": 1.53,
      "learning_rate": 1e-05,
      "loss": 0.6912,
      "step": 386
    },
    {
      "epoch": 1.53,
      "learning_rate": 1e-05,
      "loss": 0.6988,
      "step": 387
    },
    {
      "epoch": 1.54,
      "learning_rate": 1e-05,
      "loss": 0.6908,
      "step": 388
    },
    {
      "epoch": 1.54,
      "learning_rate": 1e-05,
      "loss": 0.6944,
      "step": 389
    },
    {
      "epoch": 1.55,
      "learning_rate": 1e-05,
      "loss": 0.6953,
      "step": 390
    },
    {
      "epoch": 1.55,
      "learning_rate": 1e-05,
      "loss": 0.6648,
      "step": 391
    },
    {
      "epoch": 1.55,
      "learning_rate": 1e-05,
      "loss": 0.686,
      "step": 392
    },
    {
      "epoch": 1.56,
      "learning_rate": 1e-05,
      "loss": 0.6776,
      "step": 393
    },
    {
      "epoch": 1.56,
      "learning_rate": 1e-05,
      "loss": 0.6716,
      "step": 394
    },
    {
      "epoch": 1.57,
      "learning_rate": 1e-05,
      "loss": 0.6742,
      "step": 395
    },
    {
      "epoch": 1.57,
      "learning_rate": 1e-05,
      "loss": 0.663,
      "step": 396
    },
    {
      "epoch": 1.57,
      "learning_rate": 1e-05,
      "loss": 0.6796,
      "step": 397
    },
    {
      "epoch": 1.58,
      "learning_rate": 1e-05,
      "loss": 0.6758,
      "step": 398
    },
    {
      "epoch": 1.58,
      "learning_rate": 1e-05,
      "loss": 0.6671,
      "step": 399
    },
    {
      "epoch": 1.59,
      "learning_rate": 1e-05,
      "loss": 0.6747,
      "step": 400
    },
    {
      "epoch": 1.59,
      "learning_rate": 1e-05,
      "loss": 0.6851,
      "step": 401
    },
    {
      "epoch": 1.59,
      "learning_rate": 1e-05,
      "loss": 0.672,
      "step": 402
    },
    {
      "epoch": 1.6,
      "learning_rate": 1e-05,
      "loss": 0.6563,
      "step": 403
    },
    {
      "epoch": 1.6,
      "learning_rate": 1e-05,
      "loss": 0.6671,
      "step": 404
    },
    {
      "epoch": 1.61,
      "learning_rate": 1e-05,
      "loss": 0.6748,
      "step": 405
    },
    {
      "epoch": 1.61,
      "learning_rate": 1e-05,
      "loss": 0.6986,
      "step": 406
    },
    {
      "epoch": 1.61,
      "learning_rate": 1e-05,
      "loss": 0.6935,
      "step": 407
    },
    {
      "epoch": 1.62,
      "learning_rate": 1e-05,
      "loss": 0.6781,
      "step": 408
    },
    {
      "epoch": 1.62,
      "learning_rate": 1e-05,
      "loss": 0.6774,
      "step": 409
    },
    {
      "epoch": 1.63,
      "learning_rate": 1e-05,
      "loss": 0.689,
      "step": 410
    },
    {
      "epoch": 1.63,
      "learning_rate": 1e-05,
      "loss": 0.6866,
      "step": 411
    },
    {
      "epoch": 1.63,
      "learning_rate": 1e-05,
      "loss": 0.6865,
      "step": 412
    },
    {
      "epoch": 1.64,
      "learning_rate": 1e-05,
      "loss": 0.6829,
      "step": 413
    },
    {
      "epoch": 1.64,
      "learning_rate": 1e-05,
      "loss": 0.6897,
      "step": 414
    },
    {
      "epoch": 1.65,
      "learning_rate": 1e-05,
      "loss": 0.6824,
      "step": 415
    },
    {
      "epoch": 1.65,
      "learning_rate": 1e-05,
      "loss": 0.6772,
      "step": 416
    },
    {
      "epoch": 1.65,
      "learning_rate": 1e-05,
      "loss": 0.6895,
      "step": 417
    },
    {
      "epoch": 1.66,
      "learning_rate": 1e-05,
      "loss": 0.6678,
      "step": 418
    },
    {
      "epoch": 1.66,
      "learning_rate": 1e-05,
      "loss": 0.685,
      "step": 419
    },
    {
      "epoch": 1.67,
      "learning_rate": 1e-05,
      "loss": 0.6731,
      "step": 420
    },
    {
      "epoch": 1.67,
      "learning_rate": 1e-05,
      "loss": 0.6672,
      "step": 421
    },
    {
      "epoch": 1.67,
      "learning_rate": 1e-05,
      "loss": 0.6986,
      "step": 422
    },
    {
      "epoch": 1.68,
      "learning_rate": 1e-05,
      "loss": 0.6701,
      "step": 423
    },
    {
      "epoch": 1.68,
      "learning_rate": 1e-05,
      "loss": 0.6561,
      "step": 424
    },
    {
      "epoch": 1.69,
      "learning_rate": 1e-05,
      "loss": 0.6701,
      "step": 425
    },
    {
      "epoch": 1.69,
      "learning_rate": 1e-05,
      "loss": 0.7023,
      "step": 426
    },
    {
      "epoch": 1.69,
      "learning_rate": 1e-05,
      "loss": 0.6546,
      "step": 427
    },
    {
      "epoch": 1.7,
      "learning_rate": 1e-05,
      "loss": 0.6609,
      "step": 428
    },
    {
      "epoch": 1.7,
      "learning_rate": 1e-05,
      "loss": 0.6699,
      "step": 429
    },
    {
      "epoch": 1.71,
      "learning_rate": 1e-05,
      "loss": 0.6567,
      "step": 430
    },
    {
      "epoch": 1.71,
      "learning_rate": 1e-05,
      "loss": 0.6913,
      "step": 431
    },
    {
      "epoch": 1.71,
      "learning_rate": 1e-05,
      "loss": 0.6743,
      "step": 432
    },
    {
      "epoch": 1.72,
      "learning_rate": 1e-05,
      "loss": 0.6797,
      "step": 433
    },
    {
      "epoch": 1.72,
      "learning_rate": 1e-05,
      "loss": 0.6595,
      "step": 434
    },
    {
      "epoch": 1.72,
      "learning_rate": 1e-05,
      "loss": 0.6772,
      "step": 435
    },
    {
      "epoch": 1.73,
      "learning_rate": 1e-05,
      "loss": 0.6414,
      "step": 436
    },
    {
      "epoch": 1.73,
      "learning_rate": 1e-05,
      "loss": 0.6669,
      "step": 437
    },
    {
      "epoch": 1.74,
      "learning_rate": 1e-05,
      "loss": 0.6793,
      "step": 438
    },
    {
      "epoch": 1.74,
      "learning_rate": 1e-05,
      "loss": 0.6723,
      "step": 439
    },
    {
      "epoch": 1.74,
      "learning_rate": 1e-05,
      "loss": 0.6536,
      "step": 440
    },
    {
      "epoch": 1.75,
      "learning_rate": 1e-05,
      "loss": 0.6862,
      "step": 441
    },
    {
      "epoch": 1.75,
      "learning_rate": 1e-05,
      "loss": 0.6533,
      "step": 442
    },
    {
      "epoch": 1.76,
      "learning_rate": 1e-05,
      "loss": 0.6689,
      "step": 443
    },
    {
      "epoch": 1.76,
      "learning_rate": 1e-05,
      "loss": 0.6696,
      "step": 444
    },
    {
      "epoch": 1.76,
      "learning_rate": 1e-05,
      "loss": 0.6647,
      "step": 445
    },
    {
      "epoch": 1.77,
      "learning_rate": 1e-05,
      "loss": 0.6497,
      "step": 446
    },
    {
      "epoch": 1.77,
      "learning_rate": 1e-05,
      "loss": 0.6554,
      "step": 447
    },
    {
      "epoch": 1.78,
      "learning_rate": 1e-05,
      "loss": 0.6734,
      "step": 448
    },
    {
      "epoch": 1.78,
      "learning_rate": 1e-05,
      "loss": 0.6685,
      "step": 449
    },
    {
      "epoch": 1.78,
      "learning_rate": 1e-05,
      "loss": 0.6638,
      "step": 450
    },
    {
      "epoch": 1.79,
      "learning_rate": 1e-05,
      "loss": 0.6707,
      "step": 451
    },
    {
      "epoch": 1.79,
      "learning_rate": 1e-05,
      "loss": 0.6688,
      "step": 452
    },
    {
      "epoch": 1.8,
      "learning_rate": 1e-05,
      "loss": 0.6793,
      "step": 453
    },
    {
      "epoch": 1.8,
      "learning_rate": 1e-05,
      "loss": 0.6347,
      "step": 454
    },
    {
      "epoch": 1.8,
      "learning_rate": 1e-05,
      "loss": 0.6654,
      "step": 455
    },
    {
      "epoch": 1.81,
      "learning_rate": 1e-05,
      "loss": 0.6622,
      "step": 456
    },
    {
      "epoch": 1.81,
      "learning_rate": 1e-05,
      "loss": 0.6748,
      "step": 457
    },
    {
      "epoch": 1.82,
      "learning_rate": 1e-05,
      "loss": 0.6764,
      "step": 458
    },
    {
      "epoch": 1.82,
      "learning_rate": 1e-05,
      "loss": 0.6683,
      "step": 459
    },
    {
      "epoch": 1.82,
      "learning_rate": 1e-05,
      "loss": 0.6484,
      "step": 460
    },
    {
      "epoch": 1.83,
      "learning_rate": 1e-05,
      "loss": 0.6778,
      "step": 461
    },
    {
      "epoch": 1.83,
      "learning_rate": 1e-05,
      "loss": 0.6532,
      "step": 462
    },
    {
      "epoch": 1.84,
      "learning_rate": 1e-05,
      "loss": 0.6759,
      "step": 463
    },
    {
      "epoch": 1.84,
      "learning_rate": 1e-05,
      "loss": 0.6621,
      "step": 464
    },
    {
      "epoch": 1.84,
      "learning_rate": 1e-05,
      "loss": 0.6589,
      "step": 465
    },
    {
      "epoch": 1.85,
      "learning_rate": 1e-05,
      "loss": 0.6906,
      "step": 466
    },
    {
      "epoch": 1.85,
      "learning_rate": 1e-05,
      "loss": 0.6393,
      "step": 467
    },
    {
      "epoch": 1.86,
      "learning_rate": 1e-05,
      "loss": 0.6528,
      "step": 468
    },
    {
      "epoch": 1.86,
      "learning_rate": 1e-05,
      "loss": 0.6466,
      "step": 469
    },
    {
      "epoch": 1.86,
      "learning_rate": 1e-05,
      "loss": 0.6508,
      "step": 470
    },
    {
      "epoch": 1.87,
      "learning_rate": 1e-05,
      "loss": 0.6738,
      "step": 471
    },
    {
      "epoch": 1.87,
      "learning_rate": 1e-05,
      "loss": 0.6498,
      "step": 472
    },
    {
      "epoch": 1.88,
      "learning_rate": 1e-05,
      "loss": 0.655,
      "step": 473
    },
    {
      "epoch": 1.88,
      "learning_rate": 1e-05,
      "loss": 0.656,
      "step": 474
    },
    {
      "epoch": 1.88,
      "learning_rate": 1e-05,
      "loss": 0.6491,
      "step": 475
    },
    {
      "epoch": 1.89,
      "learning_rate": 1e-05,
      "loss": 0.6548,
      "step": 476
    },
    {
      "epoch": 1.89,
      "learning_rate": 1e-05,
      "loss": 0.6684,
      "step": 477
    },
    {
      "epoch": 1.9,
      "learning_rate": 1e-05,
      "loss": 0.6573,
      "step": 478
    },
    {
      "epoch": 1.9,
      "learning_rate": 1e-05,
      "loss": 0.6654,
      "step": 479
    },
    {
      "epoch": 1.9,
      "learning_rate": 1e-05,
      "loss": 0.6682,
      "step": 480
    },
    {
      "epoch": 1.91,
      "learning_rate": 1e-05,
      "loss": 0.6757,
      "step": 481
    },
    {
      "epoch": 1.91,
      "learning_rate": 1e-05,
      "loss": 0.6535,
      "step": 482
    },
    {
      "epoch": 1.92,
      "learning_rate": 1e-05,
      "loss": 0.6383,
      "step": 483
    },
    {
      "epoch": 1.92,
      "learning_rate": 1e-05,
      "loss": 0.6593,
      "step": 484
    },
    {
      "epoch": 1.92,
      "learning_rate": 1e-05,
      "loss": 0.6485,
      "step": 485
    },
    {
      "epoch": 1.93,
      "learning_rate": 1e-05,
      "loss": 0.6603,
      "step": 486
    },
    {
      "epoch": 1.93,
      "learning_rate": 1e-05,
      "loss": 0.6545,
      "step": 487
    },
    {
      "epoch": 1.94,
      "learning_rate": 1e-05,
      "loss": 0.6673,
      "step": 488
    },
    {
      "epoch": 1.94,
      "learning_rate": 1e-05,
      "loss": 0.6606,
      "step": 489
    },
    {
      "epoch": 1.94,
      "learning_rate": 1e-05,
      "loss": 0.6649,
      "step": 490
    },
    {
      "epoch": 1.95,
      "learning_rate": 1e-05,
      "loss": 0.6498,
      "step": 491
    },
    {
      "epoch": 1.95,
      "learning_rate": 1e-05,
      "loss": 0.6688,
      "step": 492
    },
    {
      "epoch": 1.95,
      "learning_rate": 1e-05,
      "loss": 0.6723,
      "step": 493
    },
    {
      "epoch": 1.96,
      "learning_rate": 1e-05,
      "loss": 0.6407,
      "step": 494
    },
    {
      "epoch": 1.96,
      "learning_rate": 1e-05,
      "loss": 0.6391,
      "step": 495
    },
    {
      "epoch": 1.97,
      "learning_rate": 1e-05,
      "loss": 0.6502,
      "step": 496
    },
    {
      "epoch": 1.97,
      "learning_rate": 1e-05,
      "loss": 0.653,
      "step": 497
    },
    {
      "epoch": 1.97,
      "learning_rate": 1e-05,
      "loss": 0.6544,
      "step": 498
    },
    {
      "epoch": 1.98,
      "learning_rate": 1e-05,
      "loss": 0.6585,
      "step": 499
    },
    {
      "epoch": 1.98,
      "learning_rate": 1e-05,
      "loss": 0.654,
      "step": 500
    },
    {
      "epoch": 1.99,
      "learning_rate": 1e-05,
      "loss": 0.6502,
      "step": 501
    },
    {
      "epoch": 1.99,
      "learning_rate": 1e-05,
      "loss": 0.6415,
      "step": 502
    },
    {
      "epoch": 1.99,
      "learning_rate": 1e-05,
      "loss": 0.6746,
      "step": 503
    },
    {
      "epoch": 2.0,
      "learning_rate": 1e-05,
      "loss": 0.6698,
      "step": 504
    },
    {
      "epoch": 2.0,
      "learning_rate": 1e-05,
      "loss": 0.5898,
      "step": 505
    },
    {
      "epoch": 2.01,
      "learning_rate": 1e-05,
      "loss": 0.5663,
      "step": 506
    },
    {
      "epoch": 2.01,
      "learning_rate": 1e-05,
      "loss": 0.566,
      "step": 507
    },
    {
      "epoch": 2.01,
      "learning_rate": 1e-05,
      "loss": 0.5652,
      "step": 508
    },
    {
      "epoch": 2.02,
      "learning_rate": 1e-05,
      "loss": 0.5562,
      "step": 509
    },
    {
      "epoch": 2.02,
      "learning_rate": 1e-05,
      "loss": 0.5749,
      "step": 510
    },
    {
      "epoch": 2.03,
      "learning_rate": 1e-05,
      "loss": 0.5476,
      "step": 511
    },
    {
      "epoch": 2.03,
      "learning_rate": 1e-05,
      "loss": 0.5745,
      "step": 512
    },
    {
      "epoch": 2.03,
      "learning_rate": 1e-05,
      "loss": 0.5615,
      "step": 513
    },
    {
      "epoch": 2.04,
      "learning_rate": 1e-05,
      "loss": 0.5422,
      "step": 514
    },
    {
      "epoch": 2.04,
      "learning_rate": 1e-05,
      "loss": 0.5452,
      "step": 515
    },
    {
      "epoch": 2.05,
      "learning_rate": 1e-05,
      "loss": 0.575,
      "step": 516
    },
    {
      "epoch": 2.05,
      "learning_rate": 1e-05,
      "loss": 0.5655,
      "step": 517
    },
    {
      "epoch": 2.05,
      "learning_rate": 1e-05,
      "loss": 0.5441,
      "step": 518
    },
    {
      "epoch": 2.06,
      "learning_rate": 1e-05,
      "loss": 0.5576,
      "step": 519
    },
    {
      "epoch": 2.06,
      "learning_rate": 1e-05,
      "loss": 0.5571,
      "step": 520
    },
    {
      "epoch": 2.07,
      "learning_rate": 1e-05,
      "loss": 0.554,
      "step": 521
    },
    {
      "epoch": 2.07,
      "learning_rate": 1e-05,
      "loss": 0.5717,
      "step": 522
    },
    {
      "epoch": 2.07,
      "learning_rate": 1e-05,
      "loss": 0.5553,
      "step": 523
    },
    {
      "epoch": 2.08,
      "learning_rate": 1e-05,
      "loss": 0.5679,
      "step": 524
    },
    {
      "epoch": 2.08,
      "learning_rate": 1e-05,
      "loss": 0.5514,
      "step": 525
    },
    {
      "epoch": 2.09,
      "learning_rate": 1e-05,
      "loss": 0.5572,
      "step": 526
    },
    {
      "epoch": 2.09,
      "learning_rate": 1e-05,
      "loss": 0.5626,
      "step": 527
    },
    {
      "epoch": 2.09,
      "learning_rate": 1e-05,
      "loss": 0.5531,
      "step": 528
    },
    {
      "epoch": 2.1,
      "learning_rate": 1e-05,
      "loss": 0.5388,
      "step": 529
    },
    {
      "epoch": 2.1,
      "learning_rate": 1e-05,
      "loss": 0.5431,
      "step": 530
    },
    {
      "epoch": 2.11,
      "learning_rate": 1e-05,
      "loss": 0.5522,
      "step": 531
    },
    {
      "epoch": 2.11,
      "learning_rate": 1e-05,
      "loss": 0.5412,
      "step": 532
    },
    {
      "epoch": 2.11,
      "learning_rate": 1e-05,
      "loss": 0.5469,
      "step": 533
    },
    {
      "epoch": 2.12,
      "learning_rate": 1e-05,
      "loss": 0.5505,
      "step": 534
    },
    {
      "epoch": 2.12,
      "learning_rate": 1e-05,
      "loss": 0.5348,
      "step": 535
    },
    {
      "epoch": 2.13,
      "learning_rate": 1e-05,
      "loss": 0.5736,
      "step": 536
    },
    {
      "epoch": 2.13,
      "learning_rate": 1e-05,
      "loss": 0.563,
      "step": 537
    },
    {
      "epoch": 2.13,
      "learning_rate": 1e-05,
      "loss": 0.5663,
      "step": 538
    },
    {
      "epoch": 2.14,
      "learning_rate": 1e-05,
      "loss": 0.5559,
      "step": 539
    },
    {
      "epoch": 2.14,
      "learning_rate": 1e-05,
      "loss": 0.5563,
      "step": 540
    },
    {
      "epoch": 2.15,
      "learning_rate": 1e-05,
      "loss": 0.5717,
      "step": 541
    },
    {
      "epoch": 2.15,
      "learning_rate": 1e-05,
      "loss": 0.5477,
      "step": 542
    },
    {
      "epoch": 2.15,
      "learning_rate": 1e-05,
      "loss": 0.5447,
      "step": 543
    },
    {
      "epoch": 2.16,
      "learning_rate": 1e-05,
      "loss": 0.5516,
      "step": 544
    },
    {
      "epoch": 2.16,
      "learning_rate": 1e-05,
      "loss": 0.5585,
      "step": 545
    },
    {
      "epoch": 2.17,
      "learning_rate": 1e-05,
      "loss": 0.556,
      "step": 546
    },
    {
      "epoch": 2.17,
      "learning_rate": 1e-05,
      "loss": 0.5301,
      "step": 547
    },
    {
      "epoch": 2.17,
      "learning_rate": 1e-05,
      "loss": 0.5745,
      "step": 548
    },
    {
      "epoch": 2.18,
      "learning_rate": 1e-05,
      "loss": 0.5706,
      "step": 549
    },
    {
      "epoch": 2.18,
      "learning_rate": 1e-05,
      "loss": 0.5546,
      "step": 550
    },
    {
      "epoch": 2.18,
      "learning_rate": 1e-05,
      "loss": 0.5448,
      "step": 551
    },
    {
      "epoch": 2.19,
      "learning_rate": 1e-05,
      "loss": 0.5432,
      "step": 552
    },
    {
      "epoch": 2.19,
      "learning_rate": 1e-05,
      "loss": 0.5636,
      "step": 553
    },
    {
      "epoch": 2.2,
      "learning_rate": 1e-05,
      "loss": 0.5428,
      "step": 554
    },
    {
      "epoch": 2.2,
      "learning_rate": 1e-05,
      "loss": 0.5577,
      "step": 555
    },
    {
      "epoch": 2.2,
      "learning_rate": 1e-05,
      "loss": 0.5344,
      "step": 556
    },
    {
      "epoch": 2.21,
      "learning_rate": 1e-05,
      "loss": 0.5534,
      "step": 557
    },
    {
      "epoch": 2.21,
      "learning_rate": 1e-05,
      "loss": 0.5626,
      "step": 558
    },
    {
      "epoch": 2.22,
      "learning_rate": 1e-05,
      "loss": 0.5467,
      "step": 559
    },
    {
      "epoch": 2.22,
      "learning_rate": 1e-05,
      "loss": 0.5675,
      "step": 560
    },
    {
      "epoch": 2.22,
      "learning_rate": 1e-05,
      "loss": 0.5591,
      "step": 561
    },
    {
      "epoch": 2.23,
      "learning_rate": 1e-05,
      "loss": 0.5474,
      "step": 562
    },
    {
      "epoch": 2.23,
      "learning_rate": 1e-05,
      "loss": 0.5673,
      "step": 563
    },
    {
      "epoch": 2.24,
      "learning_rate": 1e-05,
      "loss": 0.5571,
      "step": 564
    },
    {
      "epoch": 2.24,
      "learning_rate": 1e-05,
      "loss": 0.548,
      "step": 565
    },
    {
      "epoch": 2.24,
      "learning_rate": 1e-05,
      "loss": 0.5444,
      "step": 566
    },
    {
      "epoch": 2.25,
      "learning_rate": 1e-05,
      "loss": 0.5581,
      "step": 567
    },
    {
      "epoch": 2.25,
      "learning_rate": 1e-05,
      "loss": 0.5576,
      "step": 568
    },
    {
      "epoch": 2.26,
      "learning_rate": 1e-05,
      "loss": 0.5431,
      "step": 569
    },
    {
      "epoch": 2.26,
      "learning_rate": 1e-05,
      "loss": 0.5393,
      "step": 570
    },
    {
      "epoch": 2.26,
      "learning_rate": 1e-05,
      "loss": 0.5619,
      "step": 571
    },
    {
      "epoch": 2.27,
      "learning_rate": 1e-05,
      "loss": 0.5674,
      "step": 572
    },
    {
      "epoch": 2.27,
      "learning_rate": 1e-05,
      "loss": 0.5584,
      "step": 573
    },
    {
      "epoch": 2.28,
      "learning_rate": 1e-05,
      "loss": 0.5458,
      "step": 574
    },
    {
      "epoch": 2.28,
      "learning_rate": 1e-05,
      "loss": 0.5539,
      "step": 575
    },
    {
      "epoch": 2.28,
      "learning_rate": 1e-05,
      "loss": 0.5533,
      "step": 576
    },
    {
      "epoch": 2.29,
      "learning_rate": 1e-05,
      "loss": 0.5536,
      "step": 577
    },
    {
      "epoch": 2.29,
      "learning_rate": 1e-05,
      "loss": 0.5232,
      "step": 578
    },
    {
      "epoch": 2.3,
      "learning_rate": 1e-05,
      "loss": 0.5315,
      "step": 579
    },
    {
      "epoch": 2.3,
      "learning_rate": 1e-05,
      "loss": 0.5875,
      "step": 580
    },
    {
      "epoch": 2.3,
      "learning_rate": 1e-05,
      "loss": 0.5412,
      "step": 581
    },
    {
      "epoch": 2.31,
      "learning_rate": 1e-05,
      "loss": 0.5564,
      "step": 582
    },
    {
      "epoch": 2.31,
      "learning_rate": 1e-05,
      "loss": 0.5443,
      "step": 583
    },
    {
      "epoch": 2.32,
      "learning_rate": 1e-05,
      "loss": 0.549,
      "step": 584
    },
    {
      "epoch": 2.32,
      "learning_rate": 1e-05,
      "loss": 0.5666,
      "step": 585
    },
    {
      "epoch": 2.32,
      "learning_rate": 1e-05,
      "loss": 0.5407,
      "step": 586
    },
    {
      "epoch": 2.33,
      "learning_rate": 1e-05,
      "loss": 0.5329,
      "step": 587
    },
    {
      "epoch": 2.33,
      "learning_rate": 1e-05,
      "loss": 0.5517,
      "step": 588
    },
    {
      "epoch": 2.34,
      "learning_rate": 1e-05,
      "loss": 0.5396,
      "step": 589
    },
    {
      "epoch": 2.34,
      "learning_rate": 1e-05,
      "loss": 0.5436,
      "step": 590
    },
    {
      "epoch": 2.34,
      "learning_rate": 1e-05,
      "loss": 0.5372,
      "step": 591
    },
    {
      "epoch": 2.35,
      "learning_rate": 1e-05,
      "loss": 0.5518,
      "step": 592
    },
    {
      "epoch": 2.35,
      "learning_rate": 1e-05,
      "loss": 0.5544,
      "step": 593
    },
    {
      "epoch": 2.36,
      "learning_rate": 1e-05,
      "loss": 0.5577,
      "step": 594
    },
    {
      "epoch": 2.36,
      "learning_rate": 1e-05,
      "loss": 0.54,
      "step": 595
    },
    {
      "epoch": 2.36,
      "learning_rate": 1e-05,
      "loss": 0.5627,
      "step": 596
    },
    {
      "epoch": 2.37,
      "learning_rate": 1e-05,
      "loss": 0.5482,
      "step": 597
    },
    {
      "epoch": 2.37,
      "learning_rate": 1e-05,
      "loss": 0.5386,
      "step": 598
    },
    {
      "epoch": 2.38,
      "learning_rate": 1e-05,
      "loss": 0.5499,
      "step": 599
    },
    {
      "epoch": 2.38,
      "learning_rate": 1e-05,
      "loss": 0.5405,
      "step": 600
    },
    {
      "epoch": 2.38,
      "learning_rate": 1e-05,
      "loss": 0.5397,
      "step": 601
    },
    {
      "epoch": 2.39,
      "learning_rate": 1e-05,
      "loss": 0.5495,
      "step": 602
    },
    {
      "epoch": 2.39,
      "learning_rate": 1e-05,
      "loss": 0.5498,
      "step": 603
    },
    {
      "epoch": 2.4,
      "learning_rate": 1e-05,
      "loss": 0.5459,
      "step": 604
    },
    {
      "epoch": 2.4,
      "learning_rate": 1e-05,
      "loss": 0.5363,
      "step": 605
    },
    {
      "epoch": 2.4,
      "learning_rate": 1e-05,
      "loss": 0.5473,
      "step": 606
    },
    {
      "epoch": 2.41,
      "learning_rate": 1e-05,
      "loss": 0.542,
      "step": 607
    },
    {
      "epoch": 2.41,
      "learning_rate": 1e-05,
      "loss": 0.5471,
      "step": 608
    },
    {
      "epoch": 2.41,
      "learning_rate": 1e-05,
      "loss": 0.5618,
      "step": 609
    },
    {
      "epoch": 2.42,
      "learning_rate": 1e-05,
      "loss": 0.5511,
      "step": 610
    },
    {
      "epoch": 2.42,
      "learning_rate": 1e-05,
      "loss": 0.5374,
      "step": 611
    },
    {
      "epoch": 2.43,
      "learning_rate": 1e-05,
      "loss": 0.5537,
      "step": 612
    },
    {
      "epoch": 2.43,
      "learning_rate": 1e-05,
      "loss": 0.553,
      "step": 613
    },
    {
      "epoch": 2.43,
      "learning_rate": 1e-05,
      "loss": 0.5595,
      "step": 614
    },
    {
      "epoch": 2.44,
      "learning_rate": 1e-05,
      "loss": 0.5381,
      "step": 615
    },
    {
      "epoch": 2.44,
      "learning_rate": 1e-05,
      "loss": 0.5456,
      "step": 616
    },
    {
      "epoch": 2.45,
      "learning_rate": 1e-05,
      "loss": 0.5441,
      "step": 617
    },
    {
      "epoch": 2.45,
      "learning_rate": 1e-05,
      "loss": 0.5294,
      "step": 618
    },
    {
      "epoch": 2.45,
      "learning_rate": 1e-05,
      "loss": 0.5503,
      "step": 619
    },
    {
      "epoch": 2.46,
      "learning_rate": 1e-05,
      "loss": 0.5405,
      "step": 620
    },
    {
      "epoch": 2.46,
      "learning_rate": 1e-05,
      "loss": 0.5574,
      "step": 621
    },
    {
      "epoch": 2.47,
      "learning_rate": 1e-05,
      "loss": 0.541,
      "step": 622
    },
    {
      "epoch": 2.47,
      "learning_rate": 1e-05,
      "loss": 0.5105,
      "step": 623
    },
    {
      "epoch": 2.47,
      "learning_rate": 1e-05,
      "loss": 0.5537,
      "step": 624
    },
    {
      "epoch": 2.48,
      "learning_rate": 1e-05,
      "loss": 0.5471,
      "step": 625
    },
    {
      "epoch": 2.48,
      "learning_rate": 1e-05,
      "loss": 0.5429,
      "step": 626
    },
    {
      "epoch": 2.49,
      "learning_rate": 1e-05,
      "loss": 0.531,
      "step": 627
    },
    {
      "epoch": 2.49,
      "learning_rate": 1e-05,
      "loss": 0.5287,
      "step": 628
    },
    {
      "epoch": 2.49,
      "learning_rate": 1e-05,
      "loss": 0.557,
      "step": 629
    },
    {
      "epoch": 2.5,
      "learning_rate": 1e-05,
      "loss": 0.5482,
      "step": 630
    },
    {
      "epoch": 2.5,
      "learning_rate": 1e-05,
      "loss": 0.5501,
      "step": 631
    },
    {
      "epoch": 2.51,
      "learning_rate": 1e-05,
      "loss": 0.554,
      "step": 632
    },
    {
      "epoch": 2.51,
      "learning_rate": 1e-05,
      "loss": 0.538,
      "step": 633
    },
    {
      "epoch": 2.51,
      "learning_rate": 1e-05,
      "loss": 0.5172,
      "step": 634
    },
    {
      "epoch": 2.52,
      "learning_rate": 1e-05,
      "loss": 0.5335,
      "step": 635
    },
    {
      "epoch": 2.52,
      "learning_rate": 1e-05,
      "loss": 0.5213,
      "step": 636
    },
    {
      "epoch": 2.53,
      "learning_rate": 1e-05,
      "loss": 0.5617,
      "step": 637
    },
    {
      "epoch": 2.53,
      "learning_rate": 1e-05,
      "loss": 0.5499,
      "step": 638
    },
    {
      "epoch": 2.53,
      "learning_rate": 1e-05,
      "loss": 0.5367,
      "step": 639
    },
    {
      "epoch": 2.54,
      "learning_rate": 1e-05,
      "loss": 0.5475,
      "step": 640
    },
    {
      "epoch": 2.54,
      "learning_rate": 1e-05,
      "loss": 0.5428,
      "step": 641
    },
    {
      "epoch": 2.55,
      "learning_rate": 1e-05,
      "loss": 0.5395,
      "step": 642
    },
    {
      "epoch": 2.55,
      "learning_rate": 1e-05,
      "loss": 0.5252,
      "step": 643
    },
    {
      "epoch": 2.55,
      "learning_rate": 1e-05,
      "loss": 0.5394,
      "step": 644
    },
    {
      "epoch": 2.56,
      "learning_rate": 1e-05,
      "loss": 0.5706,
      "step": 645
    },
    {
      "epoch": 2.56,
      "learning_rate": 1e-05,
      "loss": 0.5345,
      "step": 646
    },
    {
      "epoch": 2.57,
      "learning_rate": 1e-05,
      "loss": 0.5399,
      "step": 647
    },
    {
      "epoch": 2.57,
      "learning_rate": 1e-05,
      "loss": 0.5376,
      "step": 648
    },
    {
      "epoch": 2.57,
      "learning_rate": 1e-05,
      "loss": 0.5415,
      "step": 649
    },
    {
      "epoch": 2.58,
      "learning_rate": 1e-05,
      "loss": 0.5685,
      "step": 650
    },
    {
      "epoch": 2.58,
      "learning_rate": 1e-05,
      "loss": 0.5511,
      "step": 651
    },
    {
      "epoch": 2.59,
      "learning_rate": 1e-05,
      "loss": 0.5589,
      "step": 652
    },
    {
      "epoch": 2.59,
      "learning_rate": 1e-05,
      "loss": 0.5149,
      "step": 653
    },
    {
      "epoch": 2.59,
      "learning_rate": 1e-05,
      "loss": 0.5387,
      "step": 654
    },
    {
      "epoch": 2.6,
      "learning_rate": 1e-05,
      "loss": 0.5729,
      "step": 655
    },
    {
      "epoch": 2.6,
      "learning_rate": 1e-05,
      "loss": 0.5485,
      "step": 656
    },
    {
      "epoch": 2.61,
      "learning_rate": 1e-05,
      "loss": 0.5266,
      "step": 657
    },
    {
      "epoch": 2.61,
      "learning_rate": 1e-05,
      "loss": 0.5722,
      "step": 658
    },
    {
      "epoch": 2.61,
      "learning_rate": 1e-05,
      "loss": 0.5369,
      "step": 659
    },
    {
      "epoch": 2.62,
      "learning_rate": 1e-05,
      "loss": 0.5295,
      "step": 660
    },
    {
      "epoch": 2.62,
      "learning_rate": 1e-05,
      "loss": 0.5445,
      "step": 661
    },
    {
      "epoch": 2.63,
      "learning_rate": 1e-05,
      "loss": 0.5205,
      "step": 662
    },
    {
      "epoch": 2.63,
      "learning_rate": 1e-05,
      "loss": 0.5436,
      "step": 663
    },
    {
      "epoch": 2.63,
      "learning_rate": 1e-05,
      "loss": 0.5454,
      "step": 664
    },
    {
      "epoch": 2.64,
      "learning_rate": 1e-05,
      "loss": 0.5311,
      "step": 665
    },
    {
      "epoch": 2.64,
      "learning_rate": 1e-05,
      "loss": 0.5438,
      "step": 666
    },
    {
      "epoch": 2.64,
      "learning_rate": 1e-05,
      "loss": 0.5435,
      "step": 667
    },
    {
      "epoch": 2.65,
      "learning_rate": 1e-05,
      "loss": 0.522,
      "step": 668
    },
    {
      "epoch": 2.65,
      "learning_rate": 1e-05,
      "loss": 0.5431,
      "step": 669
    },
    {
      "epoch": 2.66,
      "learning_rate": 1e-05,
      "loss": 0.5422,
      "step": 670
    },
    {
      "epoch": 2.66,
      "learning_rate": 1e-05,
      "loss": 0.5362,
      "step": 671
    },
    {
      "epoch": 2.66,
      "learning_rate": 1e-05,
      "loss": 0.5299,
      "step": 672
    },
    {
      "epoch": 2.67,
      "learning_rate": 1e-05,
      "loss": 0.5314,
      "step": 673
    },
    {
      "epoch": 2.67,
      "learning_rate": 1e-05,
      "loss": 0.5252,
      "step": 674
    },
    {
      "epoch": 2.68,
      "learning_rate": 1e-05,
      "loss": 0.5268,
      "step": 675
    },
    {
      "epoch": 2.68,
      "learning_rate": 1e-05,
      "loss": 0.5437,
      "step": 676
    },
    {
      "epoch": 2.68,
      "learning_rate": 1e-05,
      "loss": 0.5238,
      "step": 677
    },
    {
      "epoch": 2.69,
      "learning_rate": 1e-05,
      "loss": 0.5323,
      "step": 678
    },
    {
      "epoch": 2.69,
      "learning_rate": 1e-05,
      "loss": 0.5465,
      "step": 679
    },
    {
      "epoch": 2.7,
      "learning_rate": 1e-05,
      "loss": 0.5245,
      "step": 680
    },
    {
      "epoch": 2.7,
      "learning_rate": 1e-05,
      "loss": 0.5107,
      "step": 681
    },
    {
      "epoch": 2.7,
      "learning_rate": 1e-05,
      "loss": 0.5385,
      "step": 682
    },
    {
      "epoch": 2.71,
      "learning_rate": 1e-05,
      "loss": 0.5521,
      "step": 683
    },
    {
      "epoch": 2.71,
      "learning_rate": 1e-05,
      "loss": 0.5222,
      "step": 684
    },
    {
      "epoch": 2.72,
      "learning_rate": 1e-05,
      "loss": 0.5173,
      "step": 685
    },
    {
      "epoch": 2.72,
      "learning_rate": 1e-05,
      "loss": 0.5167,
      "step": 686
    },
    {
      "epoch": 2.72,
      "learning_rate": 1e-05,
      "loss": 0.5315,
      "step": 687
    },
    {
      "epoch": 2.73,
      "learning_rate": 1e-05,
      "loss": 0.5206,
      "step": 688
    },
    {
      "epoch": 2.73,
      "learning_rate": 1e-05,
      "loss": 0.5186,
      "step": 689
    },
    {
      "epoch": 2.74,
      "learning_rate": 1e-05,
      "loss": 0.5275,
      "step": 690
    },
    {
      "epoch": 2.74,
      "learning_rate": 1e-05,
      "loss": 0.5436,
      "step": 691
    },
    {
      "epoch": 2.74,
      "learning_rate": 1e-05,
      "loss": 0.5088,
      "step": 692
    },
    {
      "epoch": 2.75,
      "learning_rate": 1e-05,
      "loss": 0.5329,
      "step": 693
    },
    {
      "epoch": 2.75,
      "learning_rate": 1e-05,
      "loss": 0.5273,
      "step": 694
    },
    {
      "epoch": 2.76,
      "learning_rate": 1e-05,
      "loss": 0.5446,
      "step": 695
    },
    {
      "epoch": 2.76,
      "learning_rate": 1e-05,
      "loss": 0.5304,
      "step": 696
    },
    {
      "epoch": 2.76,
      "learning_rate": 1e-05,
      "loss": 0.5185,
      "step": 697
    },
    {
      "epoch": 2.77,
      "learning_rate": 1e-05,
      "loss": 0.5256,
      "step": 698
    },
    {
      "epoch": 2.77,
      "learning_rate": 1e-05,
      "loss": 0.5121,
      "step": 699
    },
    {
      "epoch": 2.78,
      "learning_rate": 1e-05,
      "loss": 0.519,
      "step": 700
    },
    {
      "epoch": 2.78,
      "learning_rate": 1e-05,
      "loss": 0.5466,
      "step": 701
    },
    {
      "epoch": 2.78,
      "learning_rate": 1e-05,
      "loss": 0.5235,
      "step": 702
    },
    {
      "epoch": 2.79,
      "learning_rate": 1e-05,
      "loss": 0.5414,
      "step": 703
    },
    {
      "epoch": 2.79,
      "learning_rate": 1e-05,
      "loss": 0.5349,
      "step": 704
    },
    {
      "epoch": 2.8,
      "learning_rate": 1e-05,
      "loss": 0.528,
      "step": 705
    },
    {
      "epoch": 2.8,
      "learning_rate": 1e-05,
      "loss": 0.5402,
      "step": 706
    },
    {
      "epoch": 2.8,
      "learning_rate": 1e-05,
      "loss": 0.537,
      "step": 707
    },
    {
      "epoch": 2.81,
      "learning_rate": 1e-05,
      "loss": 0.5009,
      "step": 708
    },
    {
      "epoch": 2.81,
      "learning_rate": 1e-05,
      "loss": 0.5449,
      "step": 709
    },
    {
      "epoch": 2.82,
      "learning_rate": 1e-05,
      "loss": 0.5501,
      "step": 710
    },
    {
      "epoch": 2.82,
      "learning_rate": 1e-05,
      "loss": 0.5344,
      "step": 711
    },
    {
      "epoch": 2.82,
      "learning_rate": 1e-05,
      "loss": 0.5179,
      "step": 712
    },
    {
      "epoch": 2.83,
      "learning_rate": 1e-05,
      "loss": 0.5318,
      "step": 713
    },
    {
      "epoch": 2.83,
      "learning_rate": 1e-05,
      "loss": 0.5242,
      "step": 714
    },
    {
      "epoch": 2.84,
      "learning_rate": 1e-05,
      "loss": 0.528,
      "step": 715
    },
    {
      "epoch": 2.84,
      "learning_rate": 1e-05,
      "loss": 0.5384,
      "step": 716
    },
    {
      "epoch": 2.84,
      "learning_rate": 1e-05,
      "loss": 0.514,
      "step": 717
    },
    {
      "epoch": 2.85,
      "learning_rate": 1e-05,
      "loss": 0.5249,
      "step": 718
    },
    {
      "epoch": 2.85,
      "learning_rate": 1e-05,
      "loss": 0.5165,
      "step": 719
    },
    {
      "epoch": 2.86,
      "learning_rate": 1e-05,
      "loss": 0.5279,
      "step": 720
    },
    {
      "epoch": 2.86,
      "learning_rate": 1e-05,
      "loss": 0.5472,
      "step": 721
    },
    {
      "epoch": 2.86,
      "learning_rate": 1e-05,
      "loss": 0.5097,
      "step": 722
    },
    {
      "epoch": 2.87,
      "learning_rate": 1e-05,
      "loss": 0.5237,
      "step": 723
    },
    {
      "epoch": 2.87,
      "learning_rate": 1e-05,
      "loss": 0.511,
      "step": 724
    },
    {
      "epoch": 2.87,
      "learning_rate": 1e-05,
      "loss": 0.5065,
      "step": 725
    },
    {
      "epoch": 2.88,
      "learning_rate": 1e-05,
      "loss": 0.5167,
      "step": 726
    },
    {
      "epoch": 2.88,
      "learning_rate": 1e-05,
      "loss": 0.5237,
      "step": 727
    },
    {
      "epoch": 2.89,
      "learning_rate": 1e-05,
      "loss": 0.5323,
      "step": 728
    },
    {
      "epoch": 2.89,
      "learning_rate": 1e-05,
      "loss": 0.5087,
      "step": 729
    },
    {
      "epoch": 2.89,
      "learning_rate": 1e-05,
      "loss": 0.5005,
      "step": 730
    },
    {
      "epoch": 2.9,
      "learning_rate": 1e-05,
      "loss": 0.5233,
      "step": 731
    },
    {
      "epoch": 2.9,
      "learning_rate": 1e-05,
      "loss": 0.5199,
      "step": 732
    },
    {
      "epoch": 2.91,
      "learning_rate": 1e-05,
      "loss": 0.5423,
      "step": 733
    },
    {
      "epoch": 2.91,
      "learning_rate": 1e-05,
      "loss": 0.5275,
      "step": 734
    },
    {
      "epoch": 2.91,
      "learning_rate": 1e-05,
      "loss": 0.5285,
      "step": 735
    },
    {
      "epoch": 2.92,
      "learning_rate": 1e-05,
      "loss": 0.5046,
      "step": 736
    },
    {
      "epoch": 2.92,
      "learning_rate": 1e-05,
      "loss": 0.513,
      "step": 737
    },
    {
      "epoch": 2.93,
      "learning_rate": 1e-05,
      "loss": 0.5357,
      "step": 738
    },
    {
      "epoch": 2.93,
      "learning_rate": 1e-05,
      "loss": 0.5227,
      "step": 739
    },
    {
      "epoch": 2.93,
      "learning_rate": 1e-05,
      "loss": 0.5265,
      "step": 740
    },
    {
      "epoch": 2.94,
      "learning_rate": 1e-05,
      "loss": 0.5183,
      "step": 741
    },
    {
      "epoch": 2.94,
      "learning_rate": 1e-05,
      "loss": 0.495,
      "step": 742
    },
    {
      "epoch": 2.95,
      "learning_rate": 1e-05,
      "loss": 0.5448,
      "step": 743
    },
    {
      "epoch": 2.95,
      "learning_rate": 1e-05,
      "loss": 0.5168,
      "step": 744
    },
    {
      "epoch": 2.95,
      "learning_rate": 1e-05,
      "loss": 0.5244,
      "step": 745
    },
    {
      "epoch": 2.96,
      "learning_rate": 1e-05,
      "loss": 0.518,
      "step": 746
    },
    {
      "epoch": 2.96,
      "learning_rate": 1e-05,
      "loss": 0.5292,
      "step": 747
    },
    {
      "epoch": 2.97,
      "learning_rate": 1e-05,
      "loss": 0.5287,
      "step": 748
    },
    {
      "epoch": 2.97,
      "learning_rate": 1e-05,
      "loss": 0.5442,
      "step": 749
    },
    {
      "epoch": 2.97,
      "learning_rate": 1e-05,
      "loss": 0.5333,
      "step": 750
    },
    {
      "epoch": 2.98,
      "learning_rate": 1e-05,
      "loss": 0.5272,
      "step": 751
    },
    {
      "epoch": 2.98,
      "learning_rate": 1e-05,
      "loss": 0.5297,
      "step": 752
    },
    {
      "epoch": 2.99,
      "learning_rate": 1e-05,
      "loss": 0.502,
      "step": 753
    },
    {
      "epoch": 2.99,
      "learning_rate": 1e-05,
      "loss": 0.5218,
      "step": 754
    },
    {
      "epoch": 2.99,
      "learning_rate": 1e-05,
      "loss": 0.5203,
      "step": 755
    },
    {
      "epoch": 3.0,
      "learning_rate": 1e-05,
      "loss": 0.5295,
      "step": 756
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 756,
  "num_train_epochs": 3,
  "save_steps": 252,
  "total_flos": 9788881423564800.0,
  "trial_name": null,
  "trial_params": null
}