{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0182861240588024,
  "global_step": 710,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "learning_rate": 1e-05,
      "loss": 1.1869,
      "step": 1
    },
    {
      "epoch": 0.0,
      "learning_rate": 1e-05,
      "loss": 1.1341,
      "step": 2
    },
    {
      "epoch": 0.0,
      "learning_rate": 1e-05,
      "loss": 1.05,
      "step": 3
    },
    {
      "epoch": 0.01,
      "learning_rate": 1e-05,
      "loss": 1.1146,
      "step": 4
    },
    {
      "epoch": 0.01,
      "learning_rate": 1e-05,
      "loss": 1.1021,
      "step": 5
    },
    {
      "epoch": 0.01,
      "learning_rate": 1e-05,
      "loss": 1.081,
      "step": 6
    },
    {
      "epoch": 0.01,
      "learning_rate": 1e-05,
      "loss": 1.3247,
      "step": 7
    },
    {
      "epoch": 0.01,
      "learning_rate": 1e-05,
      "loss": 1.317,
      "step": 8
    },
    {
      "epoch": 0.01,
      "learning_rate": 1e-05,
      "loss": 1.1557,
      "step": 9
    },
    {
      "epoch": 0.01,
      "learning_rate": 1e-05,
      "loss": 1.1719,
      "step": 10
    },
    {
      "epoch": 0.02,
      "learning_rate": 1e-05,
      "loss": 1.2996,
      "step": 11
    },
    {
      "epoch": 0.02,
      "learning_rate": 1e-05,
      "loss": 1.1536,
      "step": 12
    },
    {
      "epoch": 0.02,
      "learning_rate": 1e-05,
      "loss": 1.1461,
      "step": 13
    },
    {
      "epoch": 0.02,
      "learning_rate": 1e-05,
      "loss": 1.2701,
      "step": 14
    },
    {
      "epoch": 0.02,
      "learning_rate": 1e-05,
      "loss": 1.1228,
      "step": 15
    },
    {
      "epoch": 0.02,
      "learning_rate": 1e-05,
      "loss": 1.1551,
      "step": 16
    },
    {
      "epoch": 0.02,
      "learning_rate": 1e-05,
      "loss": 1.2681,
      "step": 17
    },
    {
      "epoch": 0.03,
      "learning_rate": 1e-05,
      "loss": 1.1348,
      "step": 18
    },
    {
      "epoch": 0.03,
      "learning_rate": 1e-05,
      "loss": 1.2159,
      "step": 19
    },
    {
      "epoch": 0.03,
      "learning_rate": 1e-05,
      "loss": 1.1969,
      "step": 20
    },
    {
      "epoch": 0.03,
      "learning_rate": 1e-05,
      "loss": 1.1143,
      "step": 21
    },
    {
      "epoch": 0.03,
      "learning_rate": 1e-05,
      "loss": 1.2252,
      "step": 22
    },
    {
      "epoch": 0.03,
      "learning_rate": 1e-05,
      "loss": 1.3024,
      "step": 23
    },
    {
      "epoch": 0.03,
      "learning_rate": 1e-05,
      "loss": 1.072,
      "step": 24
    },
    {
      "epoch": 0.04,
      "learning_rate": 1e-05,
      "loss": 1.2376,
      "step": 25
    },
    {
      "epoch": 0.04,
      "learning_rate": 1e-05,
      "loss": 1.1146,
      "step": 26
    },
    {
      "epoch": 0.04,
      "learning_rate": 1e-05,
      "loss": 1.009,
      "step": 27
    },
    {
      "epoch": 0.04,
      "learning_rate": 1e-05,
      "loss": 1.0063,
      "step": 28
    },
    {
      "epoch": 0.04,
      "learning_rate": 1e-05,
      "loss": 1.0754,
      "step": 29
    },
    {
      "epoch": 0.04,
      "learning_rate": 1e-05,
      "loss": 1.42,
      "step": 30
    },
    {
      "epoch": 0.04,
      "learning_rate": 1e-05,
      "loss": 1.1205,
      "step": 31
    },
    {
      "epoch": 0.05,
      "learning_rate": 1e-05,
      "loss": 1.2763,
      "step": 32
    },
    {
      "epoch": 0.05,
      "learning_rate": 1e-05,
      "loss": 1.1647,
      "step": 33
    },
    {
      "epoch": 0.05,
      "learning_rate": 1e-05,
      "loss": 1.2768,
      "step": 34
    },
    {
      "epoch": 0.05,
      "learning_rate": 1e-05,
      "loss": 1.3174,
      "step": 35
    },
    {
      "epoch": 0.05,
      "learning_rate": 1e-05,
      "loss": 1.1804,
      "step": 36
    },
    {
      "epoch": 0.05,
      "learning_rate": 1e-05,
      "loss": 1.3405,
      "step": 37
    },
    {
      "epoch": 0.05,
      "learning_rate": 1e-05,
      "loss": 1.3263,
      "step": 38
    },
    {
      "epoch": 0.06,
      "learning_rate": 1e-05,
      "loss": 1.1688,
      "step": 39
    },
    {
      "epoch": 0.06,
      "learning_rate": 1e-05,
      "loss": 1.1376,
      "step": 40
    },
    {
      "epoch": 0.06,
      "learning_rate": 1e-05,
      "loss": 1.0468,
      "step": 41
    },
    {
      "epoch": 0.06,
      "learning_rate": 1e-05,
      "loss": 1.2798,
      "step": 42
    },
    {
      "epoch": 0.06,
      "learning_rate": 1e-05,
      "loss": 1.1004,
      "step": 43
    },
    {
      "epoch": 0.06,
      "learning_rate": 1e-05,
      "loss": 1.147,
      "step": 44
    },
    {
      "epoch": 0.06,
      "learning_rate": 1e-05,
      "loss": 1.1532,
      "step": 45
    },
    {
      "epoch": 0.07,
      "learning_rate": 1e-05,
      "loss": 1.269,
      "step": 46
    },
    {
      "epoch": 0.07,
      "learning_rate": 1e-05,
      "loss": 1.2369,
      "step": 47
    },
    {
      "epoch": 0.07,
      "learning_rate": 1e-05,
      "loss": 1.2672,
      "step": 48
    },
    {
      "epoch": 0.07,
      "learning_rate": 1e-05,
      "loss": 1.1351,
      "step": 49
    },
    {
      "epoch": 0.07,
      "learning_rate": 1e-05,
      "loss": 1.0601,
      "step": 50
    },
    {
      "epoch": 0.07,
      "learning_rate": 1e-05,
      "loss": 1.0542,
      "step": 51
    },
    {
      "epoch": 0.07,
      "learning_rate": 1e-05,
      "loss": 1.2286,
      "step": 52
    },
    {
      "epoch": 0.08,
      "learning_rate": 1e-05,
      "loss": 1.1345,
      "step": 53
    },
    {
      "epoch": 0.08,
      "learning_rate": 1e-05,
      "loss": 1.1231,
      "step": 54
    },
    {
      "epoch": 0.08,
      "learning_rate": 1e-05,
      "loss": 1.0921,
      "step": 55
    },
    {
      "epoch": 0.08,
      "learning_rate": 1e-05,
      "loss": 0.9767,
      "step": 56
    },
    {
      "epoch": 0.08,
      "learning_rate": 1e-05,
      "loss": 1.0816,
      "step": 57
    },
    {
      "epoch": 0.08,
      "learning_rate": 1e-05,
      "loss": 1.3981,
      "step": 58
    },
    {
      "epoch": 0.08,
      "learning_rate": 1e-05,
      "loss": 1.0905,
      "step": 59
    },
    {
      "epoch": 0.09,
      "learning_rate": 1e-05,
      "loss": 1.1295,
      "step": 60
    },
    {
      "epoch": 0.09,
      "learning_rate": 1e-05,
      "loss": 1.0329,
      "step": 61
    },
    {
      "epoch": 0.09,
      "learning_rate": 1e-05,
      "loss": 1.0476,
      "step": 62
    },
    {
      "epoch": 0.09,
      "learning_rate": 1e-05,
      "loss": 1.1118,
      "step": 63
    },
    {
      "epoch": 0.09,
      "learning_rate": 1e-05,
      "loss": 1.1465,
      "step": 64
    },
    {
      "epoch": 0.09,
      "learning_rate": 1e-05,
      "loss": 1.0899,
      "step": 65
    },
    {
      "epoch": 0.09,
      "learning_rate": 1e-05,
      "loss": 1.0624,
      "step": 66
    },
    {
      "epoch": 0.1,
      "learning_rate": 1e-05,
      "loss": 1.0782,
      "step": 67
    },
    {
      "epoch": 0.1,
      "learning_rate": 1e-05,
      "loss": 1.4447,
      "step": 68
    },
    {
      "epoch": 0.1,
      "learning_rate": 1e-05,
      "loss": 1.1676,
      "step": 69
    },
    {
      "epoch": 0.1,
      "learning_rate": 1e-05,
      "loss": 1.1445,
      "step": 70
    },
    {
      "epoch": 0.1,
      "learning_rate": 1e-05,
      "loss": 1.0852,
      "step": 71
    },
    {
      "epoch": 0.1,
      "learning_rate": 1e-05,
      "loss": 1.0525,
      "step": 72
    },
    {
      "epoch": 0.1,
      "learning_rate": 1e-05,
      "loss": 1.1404,
      "step": 73
    },
    {
      "epoch": 0.11,
      "learning_rate": 1e-05,
      "loss": 1.0507,
      "step": 74
    },
    {
      "epoch": 0.11,
      "learning_rate": 1e-05,
      "loss": 1.0867,
      "step": 75
    },
    {
      "epoch": 0.11,
      "learning_rate": 1e-05,
      "loss": 1.2617,
      "step": 76
    },
    {
      "epoch": 0.11,
      "learning_rate": 1e-05,
      "loss": 1.012,
      "step": 77
    },
    {
      "epoch": 0.11,
      "learning_rate": 1e-05,
      "loss": 0.9644,
      "step": 78
    },
    {
      "epoch": 0.11,
      "learning_rate": 1e-05,
      "loss": 0.9466,
      "step": 79
    },
    {
      "epoch": 0.11,
      "learning_rate": 1e-05,
      "loss": 1.1101,
      "step": 80
    },
    {
      "epoch": 0.12,
      "learning_rate": 1e-05,
      "loss": 1.1299,
      "step": 81
    },
    {
      "epoch": 0.12,
      "learning_rate": 1e-05,
      "loss": 1.0274,
      "step": 82
    },
    {
      "epoch": 0.12,
      "learning_rate": 1e-05,
      "loss": 1.1837,
      "step": 83
    },
    {
      "epoch": 0.12,
      "learning_rate": 1e-05,
      "loss": 0.9671,
      "step": 84
    },
    {
      "epoch": 0.12,
      "learning_rate": 1e-05,
      "loss": 1.2641,
      "step": 85
    },
    {
      "epoch": 0.12,
      "learning_rate": 1e-05,
      "loss": 1.1812,
      "step": 86
    },
    {
      "epoch": 0.12,
      "learning_rate": 1e-05,
      "loss": 1.0479,
      "step": 87
    },
    {
      "epoch": 0.13,
      "learning_rate": 1e-05,
      "loss": 1.1089,
      "step": 88
    },
    {
      "epoch": 0.13,
      "learning_rate": 1e-05,
      "loss": 1.1044,
      "step": 89
    },
    {
      "epoch": 0.13,
      "learning_rate": 1e-05,
      "loss": 1.2945,
      "step": 90
    },
    {
      "epoch": 0.13,
      "learning_rate": 1e-05,
      "loss": 0.966,
      "step": 91
    },
    {
      "epoch": 0.13,
      "learning_rate": 1e-05,
      "loss": 1.1648,
      "step": 92
    },
    {
      "epoch": 0.13,
      "learning_rate": 1e-05,
      "loss": 1.0313,
      "step": 93
    },
    {
      "epoch": 0.13,
      "learning_rate": 1e-05,
      "loss": 1.2299,
      "step": 94
    },
    {
      "epoch": 0.14,
      "learning_rate": 1e-05,
      "loss": 0.9963,
      "step": 95
    },
    {
      "epoch": 0.14,
      "learning_rate": 1e-05,
      "loss": 1.0783,
      "step": 96
    },
    {
      "epoch": 0.14,
      "learning_rate": 1e-05,
      "loss": 1.1466,
      "step": 97
    },
    {
      "epoch": 0.14,
      "learning_rate": 1e-05,
      "loss": 1.0261,
      "step": 98
    },
    {
      "epoch": 0.14,
      "learning_rate": 1e-05,
      "loss": 0.9892,
      "step": 99
    },
    {
      "epoch": 0.14,
      "learning_rate": 1e-05,
      "loss": 1.4327,
      "step": 100
    },
    {
      "epoch": 0.14,
      "learning_rate": 1e-05,
      "loss": 1.1191,
      "step": 101
    },
    {
      "epoch": 0.15,
      "learning_rate": 1e-05,
      "loss": 1.1687,
      "step": 102
    },
    {
      "epoch": 0.15,
      "learning_rate": 1e-05,
      "loss": 1.0425,
      "step": 103
    },
    {
      "epoch": 0.15,
      "learning_rate": 1e-05,
      "loss": 1.0478,
      "step": 104
    },
    {
      "epoch": 0.15,
      "learning_rate": 1e-05,
      "loss": 1.1116,
      "step": 105
    },
    {
      "epoch": 0.15,
      "learning_rate": 1e-05,
      "loss": 1.1595,
      "step": 106
    },
    {
      "epoch": 0.15,
      "learning_rate": 1e-05,
      "loss": 1.1693,
      "step": 107
    },
    {
      "epoch": 0.15,
      "learning_rate": 1e-05,
      "loss": 1.2278,
      "step": 108
    },
    {
      "epoch": 0.16,
      "learning_rate": 1e-05,
      "loss": 1.0907,
      "step": 109
    },
    {
      "epoch": 0.16,
      "learning_rate": 1e-05,
      "loss": 1.1872,
      "step": 110
    },
    {
      "epoch": 0.16,
      "learning_rate": 1e-05,
      "loss": 1.0144,
      "step": 111
    },
    {
      "epoch": 0.16,
      "learning_rate": 1e-05,
      "loss": 1.0497,
      "step": 112
    },
    {
      "epoch": 0.16,
      "learning_rate": 1e-05,
      "loss": 1.1405,
      "step": 113
    },
    {
      "epoch": 0.16,
      "learning_rate": 1e-05,
      "loss": 1.0741,
      "step": 114
    },
    {
      "epoch": 0.16,
      "learning_rate": 1e-05,
      "loss": 0.9611,
      "step": 115
    },
    {
      "epoch": 0.17,
      "learning_rate": 1e-05,
      "loss": 1.146,
      "step": 116
    },
    {
      "epoch": 0.17,
      "learning_rate": 1e-05,
      "loss": 1.2032,
      "step": 117
    },
    {
      "epoch": 0.17,
      "learning_rate": 1e-05,
      "loss": 1.0621,
      "step": 118
    },
    {
      "epoch": 0.17,
      "learning_rate": 1e-05,
      "loss": 1.0425,
      "step": 119
    },
    {
      "epoch": 0.17,
      "learning_rate": 1e-05,
      "loss": 1.2721,
      "step": 120
    },
    {
      "epoch": 0.17,
      "learning_rate": 1e-05,
      "loss": 1.1062,
      "step": 121
    },
    {
      "epoch": 0.17,
      "learning_rate": 1e-05,
      "loss": 0.9715,
      "step": 122
    },
    {
      "epoch": 0.18,
      "learning_rate": 1e-05,
      "loss": 1.0765,
      "step": 123
    },
    {
      "epoch": 0.18,
      "learning_rate": 1e-05,
      "loss": 1.122,
      "step": 124
    },
    {
      "epoch": 0.18,
      "learning_rate": 1e-05,
      "loss": 1.1593,
      "step": 125
    },
    {
      "epoch": 0.18,
      "learning_rate": 1e-05,
      "loss": 1.0108,
      "step": 126
    },
    {
      "epoch": 0.18,
      "learning_rate": 1e-05,
      "loss": 1.0072,
      "step": 127
    },
    {
      "epoch": 0.18,
      "learning_rate": 1e-05,
      "loss": 1.0081,
      "step": 128
    },
    {
      "epoch": 0.19,
      "learning_rate": 1e-05,
      "loss": 1.1256,
      "step": 129
    },
    {
      "epoch": 0.19,
      "learning_rate": 1e-05,
      "loss": 1.2929,
      "step": 130
    },
    {
      "epoch": 0.19,
      "learning_rate": 1e-05,
      "loss": 1.1577,
      "step": 131
    },
    {
      "epoch": 0.19,
      "learning_rate": 1e-05,
      "loss": 1.0326,
      "step": 132
    },
    {
      "epoch": 0.19,
      "learning_rate": 1e-05,
      "loss": 1.2102,
      "step": 133
    },
    {
      "epoch": 0.19,
      "learning_rate": 1e-05,
      "loss": 1.0872,
      "step": 134
    },
    {
      "epoch": 0.19,
      "learning_rate": 1e-05,
      "loss": 1.1568,
      "step": 135
    },
    {
      "epoch": 0.2,
      "learning_rate": 1e-05,
      "loss": 1.1225,
      "step": 136
    },
    {
      "epoch": 0.2,
      "learning_rate": 1e-05,
      "loss": 1.1538,
      "step": 137
    },
    {
      "epoch": 0.2,
      "learning_rate": 1e-05,
      "loss": 1.0996,
      "step": 138
    },
    {
      "epoch": 0.2,
      "learning_rate": 1e-05,
      "loss": 0.9808,
      "step": 139
    },
    {
      "epoch": 0.2,
      "learning_rate": 1e-05,
      "loss": 1.12,
      "step": 140
    },
    {
      "epoch": 0.2,
      "learning_rate": 1e-05,
      "loss": 1.1091,
      "step": 141
    },
    {
      "epoch": 0.2,
      "learning_rate": 1e-05,
      "loss": 1.2654,
      "step": 142
    },
    {
      "epoch": 0.21,
      "learning_rate": 1e-05,
      "loss": 1.2766,
      "step": 143
    },
    {
      "epoch": 0.21,
      "learning_rate": 1e-05,
      "loss": 1.2188,
      "step": 144
    },
    {
      "epoch": 0.21,
      "learning_rate": 1e-05,
      "loss": 1.1316,
      "step": 145
    },
    {
      "epoch": 0.21,
      "learning_rate": 1e-05,
      "loss": 1.188,
      "step": 146
    },
    {
      "epoch": 0.21,
      "learning_rate": 1e-05,
      "loss": 1.1192,
      "step": 147
    },
    {
      "epoch": 0.21,
      "learning_rate": 1e-05,
      "loss": 1.1439,
      "step": 148
    },
    {
      "epoch": 0.21,
      "learning_rate": 1e-05,
      "loss": 1.0083,
      "step": 149
    },
    {
      "epoch": 0.22,
      "learning_rate": 1e-05,
      "loss": 1.0744,
      "step": 150
    },
    {
      "epoch": 0.22,
      "learning_rate": 1e-05,
      "loss": 1.1448,
      "step": 151
    },
    {
      "epoch": 0.22,
      "learning_rate": 1e-05,
      "loss": 1.0624,
      "step": 152
    },
    {
      "epoch": 0.22,
      "learning_rate": 1e-05,
      "loss": 1.0234,
      "step": 153
    },
    {
      "epoch": 0.22,
      "learning_rate": 1e-05,
      "loss": 0.9977,
      "step": 154
    },
    {
      "epoch": 0.22,
      "learning_rate": 1e-05,
      "loss": 1.0357,
      "step": 155
    },
    {
      "epoch": 0.22,
      "learning_rate": 1e-05,
      "loss": 1.1724,
      "step": 156
    },
    {
      "epoch": 0.23,
      "learning_rate": 1e-05,
      "loss": 1.0602,
      "step": 157
    },
    {
      "epoch": 0.23,
      "learning_rate": 1e-05,
      "loss": 1.1869,
      "step": 158
    },
    {
      "epoch": 0.23,
      "learning_rate": 1e-05,
      "loss": 0.9965,
      "step": 159
    },
    {
      "epoch": 0.23,
      "learning_rate": 1e-05,
      "loss": 1.1525,
      "step": 160
    },
    {
      "epoch": 0.23,
      "learning_rate": 1e-05,
      "loss": 1.0106,
      "step": 161
    },
    {
      "epoch": 0.23,
      "learning_rate": 1e-05,
      "loss": 1.0163,
      "step": 162
    },
    {
      "epoch": 0.23,
      "learning_rate": 1e-05,
      "loss": 1.0923,
      "step": 163
    },
    {
      "epoch": 0.24,
      "learning_rate": 1e-05,
      "loss": 1.1984,
      "step": 164
    },
    {
      "epoch": 0.24,
      "learning_rate": 1e-05,
      "loss": 1.0221,
      "step": 165
    },
    {
      "epoch": 0.24,
      "learning_rate": 1e-05,
      "loss": 1.0959,
      "step": 166
    },
    {
      "epoch": 0.24,
      "learning_rate": 1e-05,
      "loss": 1.0198,
      "step": 167
    },
    {
      "epoch": 0.24,
      "learning_rate": 1e-05,
      "loss": 0.9988,
      "step": 168
    },
    {
      "epoch": 0.24,
      "learning_rate": 1e-05,
      "loss": 1.1058,
      "step": 169
    },
    {
      "epoch": 0.24,
      "learning_rate": 1e-05,
      "loss": 0.9617,
      "step": 170
    },
    {
      "epoch": 0.25,
      "learning_rate": 1e-05,
      "loss": 0.9187,
      "step": 171
    },
    {
      "epoch": 0.25,
      "learning_rate": 1e-05,
      "loss": 1.0743,
      "step": 172
    },
    {
      "epoch": 0.25,
      "learning_rate": 1e-05,
      "loss": 0.9684,
      "step": 173
    },
    {
      "epoch": 0.25,
      "learning_rate": 1e-05,
      "loss": 1.0764,
      "step": 174
    },
    {
      "epoch": 0.25,
      "learning_rate": 1e-05,
      "loss": 1.0065,
      "step": 175
    },
    {
      "epoch": 0.25,
      "learning_rate": 1e-05,
      "loss": 1.064,
      "step": 176
    },
    {
      "epoch": 0.25,
      "learning_rate": 1e-05,
      "loss": 0.9899,
      "step": 177
    },
    {
      "epoch": 0.26,
      "learning_rate": 1e-05,
      "loss": 1.1346,
      "step": 178
    },
    {
      "epoch": 0.26,
      "learning_rate": 1e-05,
      "loss": 1.0037,
      "step": 179
    },
    {
      "epoch": 0.26,
      "learning_rate": 1e-05,
      "loss": 0.9849,
      "step": 180
    },
    {
      "epoch": 0.26,
      "learning_rate": 1e-05,
      "loss": 1.0597,
      "step": 181
    },
    {
      "epoch": 0.26,
      "learning_rate": 1e-05,
      "loss": 1.0869,
      "step": 182
    },
    {
      "epoch": 0.26,
      "learning_rate": 1e-05,
      "loss": 1.1165,
      "step": 183
    },
    {
      "epoch": 0.26,
      "learning_rate": 1e-05,
      "loss": 1.1071,
      "step": 184
    },
    {
      "epoch": 0.27,
      "learning_rate": 1e-05,
      "loss": 1.1435,
      "step": 185
    },
    {
      "epoch": 0.27,
      "learning_rate": 1e-05,
      "loss": 0.9519,
      "step": 186
    },
    {
      "epoch": 0.27,
      "learning_rate": 1e-05,
      "loss": 1.0177,
      "step": 187
    },
    {
      "epoch": 0.27,
      "learning_rate": 1e-05,
      "loss": 1.0496,
      "step": 188
    },
    {
      "epoch": 0.27,
      "learning_rate": 1e-05,
      "loss": 1.0887,
      "step": 189
    },
    {
      "epoch": 0.27,
      "learning_rate": 1e-05,
      "loss": 1.026,
      "step": 190
    },
    {
      "epoch": 0.27,
      "learning_rate": 1e-05,
      "loss": 0.9915,
      "step": 191
    },
    {
      "epoch": 0.28,
      "learning_rate": 1e-05,
      "loss": 1.0711,
      "step": 192
    },
    {
      "epoch": 0.28,
      "learning_rate": 1e-05,
      "loss": 1.0559,
      "step": 193
    },
    {
      "epoch": 0.28,
      "learning_rate": 1e-05,
      "loss": 1.0623,
      "step": 194
    },
    {
      "epoch": 0.28,
      "learning_rate": 1e-05,
      "loss": 1.2179,
      "step": 195
    },
    {
      "epoch": 0.28,
      "learning_rate": 1e-05,
      "loss": 1.0445,
      "step": 196
    },
    {
      "epoch": 0.28,
      "learning_rate": 1e-05,
      "loss": 1.0342,
      "step": 197
    },
    {
      "epoch": 0.28,
      "learning_rate": 1e-05,
      "loss": 0.9772,
      "step": 198
    },
    {
      "epoch": 0.29,
      "learning_rate": 1e-05,
      "loss": 1.0526,
      "step": 199
    },
    {
      "epoch": 0.29,
      "learning_rate": 1e-05,
      "loss": 1.0766,
      "step": 200
    },
    {
      "epoch": 0.29,
      "learning_rate": 1e-05,
      "loss": 1.1646,
      "step": 201
    },
    {
      "epoch": 0.29,
      "learning_rate": 1e-05,
      "loss": 1.0257,
      "step": 202
    },
    {
      "epoch": 0.29,
      "learning_rate": 1e-05,
      "loss": 1.1115,
      "step": 203
    },
    {
      "epoch": 0.29,
      "learning_rate": 1e-05,
      "loss": 0.9315,
      "step": 204
    },
    {
      "epoch": 0.29,
      "learning_rate": 1e-05,
      "loss": 1.2554,
      "step": 205
    },
    {
      "epoch": 0.3,
      "learning_rate": 1e-05,
      "loss": 0.9923,
      "step": 206
    },
    {
      "epoch": 0.3,
      "learning_rate": 1e-05,
      "loss": 1.0049,
      "step": 207
    },
    {
      "epoch": 0.3,
      "learning_rate": 1e-05,
      "loss": 1.1257,
      "step": 208
    },
    {
      "epoch": 0.3,
      "learning_rate": 1e-05,
      "loss": 1.0426,
      "step": 209
    },
    {
      "epoch": 0.3,
      "learning_rate": 1e-05,
      "loss": 1.1272,
      "step": 210
    },
    {
      "epoch": 0.3,
      "learning_rate": 1e-05,
      "loss": 1.2052,
      "step": 211
    },
    {
      "epoch": 0.3,
      "learning_rate": 1e-05,
      "loss": 1.1267,
      "step": 212
    },
    {
      "epoch": 0.31,
      "learning_rate": 1e-05,
      "loss": 1.1181,
      "step": 213
    },
    {
      "epoch": 0.31,
      "learning_rate": 1e-05,
      "loss": 1.144,
      "step": 214
    },
    {
      "epoch": 0.31,
      "learning_rate": 1e-05,
      "loss": 1.0262,
      "step": 215
    },
    {
      "epoch": 0.31,
      "learning_rate": 1e-05,
      "loss": 1.0533,
      "step": 216
    },
    {
      "epoch": 0.31,
      "learning_rate": 1e-05,
      "loss": 1.0561,
      "step": 217
    },
    {
      "epoch": 0.31,
      "learning_rate": 1e-05,
      "loss": 1.0717,
      "step": 218
    },
    {
      "epoch": 0.31,
      "learning_rate": 1e-05,
      "loss": 1.144,
      "step": 219
    },
    {
      "epoch": 0.32,
      "learning_rate": 1e-05,
      "loss": 1.1898,
      "step": 220
    },
    {
      "epoch": 0.32,
      "learning_rate": 1e-05,
      "loss": 1.0609,
      "step": 221
    },
    {
      "epoch": 0.32,
      "learning_rate": 1e-05,
      "loss": 1.1268,
      "step": 222
    },
    {
      "epoch": 0.32,
      "learning_rate": 1e-05,
      "loss": 1.0931,
      "step": 223
    },
    {
      "epoch": 0.32,
      "learning_rate": 1e-05,
      "loss": 0.9285,
      "step": 224
    },
    {
      "epoch": 0.32,
      "learning_rate": 1e-05,
      "loss": 1.0671,
      "step": 225
    },
    {
      "epoch": 0.32,
      "learning_rate": 1e-05,
      "loss": 1.0451,
      "step": 226
    },
    {
      "epoch": 0.33,
      "learning_rate": 1e-05,
      "loss": 1.0526,
      "step": 227
    },
    {
      "epoch": 0.33,
      "learning_rate": 1e-05,
      "loss": 1.0778,
      "step": 228
    },
    {
      "epoch": 0.33,
      "learning_rate": 1e-05,
      "loss": 1.032,
      "step": 229
    },
    {
      "epoch": 0.33,
      "learning_rate": 1e-05,
      "loss": 1.0474,
      "step": 230
    },
    {
      "epoch": 0.33,
      "learning_rate": 1e-05,
      "loss": 1.0434,
      "step": 231
    },
    {
      "epoch": 0.33,
      "learning_rate": 1e-05,
      "loss": 0.9927,
      "step": 232
    },
    {
      "epoch": 0.33,
      "learning_rate": 1e-05,
      "loss": 1.0242,
      "step": 233
    },
    {
      "epoch": 0.34,
      "learning_rate": 1e-05,
      "loss": 0.9503,
      "step": 234
    },
    {
      "epoch": 0.34,
      "learning_rate": 1e-05,
      "loss": 1.057,
      "step": 235
    },
    {
      "epoch": 0.34,
      "learning_rate": 1e-05,
      "loss": 1.0104,
      "step": 236
    },
    {
      "epoch": 0.34,
      "learning_rate": 1e-05,
      "loss": 1.0534,
      "step": 237
    },
    {
      "epoch": 0.34,
      "learning_rate": 1e-05,
      "loss": 1.09,
      "step": 238
    },
    {
      "epoch": 0.34,
      "learning_rate": 1e-05,
      "loss": 1.0395,
      "step": 239
    },
    {
      "epoch": 0.34,
      "learning_rate": 1e-05,
      "loss": 1.1476,
      "step": 240
    },
    {
      "epoch": 0.35,
      "learning_rate": 1e-05,
      "loss": 1.1163,
      "step": 241
    },
    {
      "epoch": 0.35,
      "learning_rate": 1e-05,
      "loss": 0.9734,
      "step": 242
    },
    {
      "epoch": 0.35,
      "learning_rate": 1e-05,
      "loss": 1.0632,
      "step": 243
    },
    {
      "epoch": 0.35,
      "learning_rate": 1e-05,
      "loss": 1.148,
      "step": 244
    },
    {
      "epoch": 0.35,
      "learning_rate": 1e-05,
      "loss": 1.0293,
      "step": 245
    },
    {
      "epoch": 0.35,
      "learning_rate": 1e-05,
      "loss": 1.1969,
      "step": 246
    },
    {
      "epoch": 0.35,
      "learning_rate": 1e-05,
      "loss": 1.0632,
      "step": 247
    },
    {
      "epoch": 0.36,
      "learning_rate": 1e-05,
      "loss": 1.1989,
      "step": 248
    },
    {
      "epoch": 0.36,
      "learning_rate": 1e-05,
      "loss": 0.9771,
      "step": 249
    },
    {
      "epoch": 0.36,
      "learning_rate": 1e-05,
      "loss": 1.0041,
      "step": 250
    },
    {
      "epoch": 0.36,
      "learning_rate": 1e-05,
      "loss": 1.03,
      "step": 251
    },
    {
      "epoch": 0.36,
      "learning_rate": 1e-05,
      "loss": 1.2347,
      "step": 252
    },
    {
      "epoch": 0.36,
      "learning_rate": 1e-05,
      "loss": 1.0142,
      "step": 253
    },
    {
      "epoch": 0.36,
      "learning_rate": 1e-05,
      "loss": 1.0632,
      "step": 254
    },
    {
      "epoch": 0.37,
      "learning_rate": 1e-05,
      "loss": 1.0263,
      "step": 255
    },
    {
      "epoch": 0.37,
      "learning_rate": 1e-05,
      "loss": 1.13,
      "step": 256
    },
    {
      "epoch": 0.37,
      "learning_rate": 1e-05,
      "loss": 1.1843,
      "step": 257
    },
    {
      "epoch": 0.37,
      "learning_rate": 1e-05,
      "loss": 0.9877,
      "step": 258
    },
    {
      "epoch": 0.37,
      "learning_rate": 1e-05,
      "loss": 0.9939,
      "step": 259
    },
    {
      "epoch": 0.37,
      "learning_rate": 1e-05,
      "loss": 1.1481,
      "step": 260
    },
    {
      "epoch": 0.37,
      "learning_rate": 1e-05,
      "loss": 1.0141,
      "step": 261
    },
    {
      "epoch": 0.38,
      "learning_rate": 1e-05,
      "loss": 1.1253,
      "step": 262
    },
    {
      "epoch": 0.38,
      "learning_rate": 1e-05,
      "loss": 1.0921,
      "step": 263
    },
    {
      "epoch": 0.38,
      "learning_rate": 1e-05,
      "loss": 0.9119,
      "step": 264
    },
    {
      "epoch": 0.38,
      "learning_rate": 1e-05,
      "loss": 1.1677,
      "step": 265
    },
    {
      "epoch": 0.38,
      "learning_rate": 1e-05,
      "loss": 1.1709,
      "step": 266
    },
    {
      "epoch": 0.38,
      "learning_rate": 1e-05,
      "loss": 1.179,
      "step": 267
    },
    {
      "epoch": 0.38,
      "learning_rate": 1e-05,
      "loss": 1.108,
      "step": 268
    },
    {
      "epoch": 0.39,
      "learning_rate": 1e-05,
      "loss": 1.1107,
      "step": 269
    },
    {
      "epoch": 0.39,
      "learning_rate": 1e-05,
      "loss": 1.0997,
      "step": 270
    },
    {
      "epoch": 0.39,
      "learning_rate": 1e-05,
      "loss": 1.1187,
      "step": 271
    },
    {
      "epoch": 0.39,
      "learning_rate": 1e-05,
      "loss": 1.0602,
      "step": 272
    },
    {
      "epoch": 0.39,
      "learning_rate": 1e-05,
      "loss": 1.0638,
      "step": 273
    },
    {
      "epoch": 0.39,
      "learning_rate": 1e-05,
      "loss": 1.0788,
      "step": 274
    },
    {
      "epoch": 0.39,
      "learning_rate": 1e-05,
      "loss": 1.0169,
      "step": 275
    },
    {
      "epoch": 0.4,
      "learning_rate": 1e-05,
      "loss": 1.1407,
      "step": 276
    },
    {
      "epoch": 0.4,
      "learning_rate": 1e-05,
      "loss": 1.0474,
      "step": 277
    },
    {
      "epoch": 0.4,
      "learning_rate": 1e-05,
      "loss": 1.1492,
      "step": 278
    },
    {
      "epoch": 0.4,
      "learning_rate": 1e-05,
      "loss": 1.1968,
      "step": 279
    },
    {
      "epoch": 0.4,
      "learning_rate": 1e-05,
      "loss": 1.1165,
      "step": 280
    },
    {
      "epoch": 0.4,
      "learning_rate": 1e-05,
      "loss": 1.1368,
      "step": 281
    },
    {
      "epoch": 0.4,
      "learning_rate": 1e-05,
      "loss": 1.0147,
      "step": 282
    },
    {
      "epoch": 0.41,
      "learning_rate": 1e-05,
      "loss": 1.0213,
      "step": 283
    },
    {
      "epoch": 0.41,
      "learning_rate": 1e-05,
      "loss": 1.1919,
      "step": 284
    },
    {
      "epoch": 0.41,
      "learning_rate": 1e-05,
      "loss": 1.0587,
      "step": 285
    },
    {
      "epoch": 0.41,
      "learning_rate": 1e-05,
      "loss": 1.1057,
      "step": 286
    },
    {
      "epoch": 0.41,
      "learning_rate": 1e-05,
      "loss": 1.1378,
      "step": 287
    },
    {
      "epoch": 0.41,
      "learning_rate": 1e-05,
      "loss": 1.069,
      "step": 288
    },
    {
      "epoch": 0.41,
      "learning_rate": 1e-05,
      "loss": 1.1614,
      "step": 289
    },
    {
      "epoch": 0.42,
      "learning_rate": 1e-05,
      "loss": 0.9797,
      "step": 290
    },
    {
      "epoch": 0.42,
      "learning_rate": 1e-05,
      "loss": 0.9816,
      "step": 291
    },
    {
      "epoch": 0.42,
      "learning_rate": 1e-05,
      "loss": 1.1907,
      "step": 292
    },
    {
      "epoch": 0.42,
      "learning_rate": 1e-05,
      "loss": 1.0368,
      "step": 293
    },
    {
      "epoch": 0.42,
      "learning_rate": 1e-05,
      "loss": 1.1285,
      "step": 294
    },
    {
      "epoch": 0.42,
      "learning_rate": 1e-05,
      "loss": 1.0124,
      "step": 295
    },
    {
      "epoch": 0.42,
      "learning_rate": 1e-05,
      "loss": 1.1616,
      "step": 296
    },
    {
      "epoch": 0.43,
      "learning_rate": 1e-05,
      "loss": 1.3592,
      "step": 297
    },
    {
      "epoch": 0.43,
      "learning_rate": 1e-05,
      "loss": 1.1724,
      "step": 298
    },
    {
      "epoch": 0.43,
      "learning_rate": 1e-05,
      "loss": 1.1562,
      "step": 299
    },
    {
      "epoch": 0.43,
      "learning_rate": 1e-05,
      "loss": 1.0077,
      "step": 300
    },
    {
      "epoch": 0.43,
      "learning_rate": 1e-05,
      "loss": 1.0588,
      "step": 301
    },
    {
      "epoch": 0.43,
      "learning_rate": 1e-05,
      "loss": 1.1677,
      "step": 302
    },
    {
      "epoch": 0.43,
      "learning_rate": 1e-05,
      "loss": 1.0943,
      "step": 303
    },
    {
      "epoch": 0.44,
      "learning_rate": 1e-05,
      "loss": 1.0741,
      "step": 304
    },
    {
      "epoch": 0.44,
      "learning_rate": 1e-05,
      "loss": 1.1019,
      "step": 305
    },
    {
      "epoch": 0.44,
      "learning_rate": 1e-05,
      "loss": 1.1682,
      "step": 306
    },
    {
      "epoch": 0.44,
      "learning_rate": 1e-05,
      "loss": 0.9391,
      "step": 307
    },
    {
      "epoch": 0.44,
      "learning_rate": 1e-05,
      "loss": 1.0799,
      "step": 308
    },
    {
      "epoch": 0.44,
      "learning_rate": 1e-05,
      "loss": 1.0364,
      "step": 309
    },
    {
      "epoch": 0.44,
      "learning_rate": 1e-05,
      "loss": 1.016,
      "step": 310
    },
    {
      "epoch": 0.45,
      "learning_rate": 1e-05,
      "loss": 0.97,
      "step": 311
    },
    {
      "epoch": 0.45,
      "learning_rate": 1e-05,
      "loss": 0.9956,
      "step": 312
    },
    {
      "epoch": 0.45,
      "learning_rate": 1e-05,
      "loss": 1.2104,
      "step": 313
    },
    {
      "epoch": 0.45,
      "learning_rate": 1e-05,
      "loss": 1.0227,
      "step": 314
    },
    {
      "epoch": 0.45,
      "learning_rate": 1e-05,
      "loss": 0.9458,
      "step": 315
    },
    {
      "epoch": 0.45,
      "learning_rate": 1e-05,
      "loss": 0.984,
      "step": 316
    },
    {
      "epoch": 0.45,
      "learning_rate": 1e-05,
      "loss": 1.0118,
      "step": 317
    },
    {
      "epoch": 0.46,
      "learning_rate": 1e-05,
      "loss": 1.0238,
      "step": 318
    },
    {
      "epoch": 0.46,
      "learning_rate": 1e-05,
      "loss": 0.8901,
      "step": 319
    },
    {
      "epoch": 0.46,
      "learning_rate": 1e-05,
      "loss": 1.1406,
      "step": 320
    },
    {
      "epoch": 0.46,
      "learning_rate": 1e-05,
      "loss": 1.2732,
      "step": 321
    },
    {
      "epoch": 0.46,
      "learning_rate": 1e-05,
      "loss": 0.8621,
      "step": 322
    },
    {
      "epoch": 0.46,
      "learning_rate": 1e-05,
      "loss": 0.9699,
      "step": 323
    },
    {
      "epoch": 0.46,
      "learning_rate": 1e-05,
      "loss": 1.0498,
      "step": 324
    },
    {
      "epoch": 0.47,
      "learning_rate": 1e-05,
      "loss": 1.0455,
      "step": 325
    },
    {
      "epoch": 0.47,
      "learning_rate": 1e-05,
      "loss": 1.2048,
      "step": 326
    },
    {
      "epoch": 0.47,
      "learning_rate": 1e-05,
      "loss": 1.1452,
      "step": 327
    },
    {
      "epoch": 0.47,
      "learning_rate": 1e-05,
      "loss": 1.024,
      "step": 328
    },
    {
      "epoch": 0.47,
      "learning_rate": 1e-05,
      "loss": 1.0851,
      "step": 329
    },
    {
      "epoch": 0.47,
      "learning_rate": 1e-05,
      "loss": 1.0247,
      "step": 330
    },
    {
      "epoch": 0.47,
      "learning_rate": 1e-05,
      "loss": 0.9912,
      "step": 331
    },
    {
      "epoch": 0.48,
      "learning_rate": 1e-05,
      "loss": 0.9362,
      "step": 332
    },
    {
      "epoch": 0.48,
      "learning_rate": 1e-05,
      "loss": 1.0321,
      "step": 333
    },
    {
      "epoch": 0.48,
      "learning_rate": 1e-05,
      "loss": 1.0211,
      "step": 334
    },
    {
      "epoch": 0.48,
      "learning_rate": 1e-05,
      "loss": 1.0143,
      "step": 335
    },
    {
      "epoch": 0.48,
      "learning_rate": 1e-05,
      "loss": 1.0649,
      "step": 336
    },
    {
      "epoch": 0.48,
      "learning_rate": 1e-05,
      "loss": 1.0055,
      "step": 337
    },
    {
      "epoch": 0.48,
      "learning_rate": 1e-05,
      "loss": 0.9968,
      "step": 338
    },
    {
      "epoch": 0.49,
      "learning_rate": 1e-05,
      "loss": 0.9522,
      "step": 339
    },
    {
      "epoch": 0.49,
      "learning_rate": 1e-05,
      "loss": 0.9609,
      "step": 340
    },
    {
      "epoch": 0.49,
      "learning_rate": 1e-05,
      "loss": 1.0265,
      "step": 341
    },
    {
      "epoch": 0.49,
      "learning_rate": 1e-05,
      "loss": 1.1408,
      "step": 342
    },
    {
      "epoch": 0.49,
      "learning_rate": 1e-05,
      "loss": 1.1828,
      "step": 343
    },
    {
      "epoch": 0.49,
      "learning_rate": 1e-05,
      "loss": 1.0147,
      "step": 344
    },
    {
      "epoch": 0.49,
      "learning_rate": 1e-05,
      "loss": 1.0865,
      "step": 345
    },
    {
      "epoch": 0.5,
      "learning_rate": 1e-05,
      "loss": 1.1806,
      "step": 346
    },
    {
      "epoch": 0.5,
      "learning_rate": 1e-05,
      "loss": 0.9917,
      "step": 347
    },
    {
      "epoch": 0.5,
      "learning_rate": 1e-05,
      "loss": 0.9569,
      "step": 348
    },
    {
      "epoch": 0.5,
      "learning_rate": 1e-05,
      "loss": 1.1141,
      "step": 349
    },
    {
      "epoch": 0.5,
      "learning_rate": 1e-05,
      "loss": 1.0834,
      "step": 350
    },
    {
      "epoch": 0.5,
      "learning_rate": 1e-05,
      "loss": 1.043,
      "step": 351
    },
    {
      "epoch": 0.5,
      "learning_rate": 1e-05,
      "loss": 1.206,
      "step": 352
    },
    {
      "epoch": 0.51,
      "learning_rate": 1e-05,
      "loss": 1.1677,
      "step": 353
    },
    {
      "epoch": 0.51,
      "learning_rate": 1e-05,
      "loss": 0.9864,
      "step": 354
    },
    {
      "epoch": 0.51,
      "learning_rate": 1e-05,
      "loss": 1.044,
      "step": 355
    },
    {
      "epoch": 0.51,
      "learning_rate": 1e-05,
      "loss": 1.1659,
      "step": 356
    },
    {
      "epoch": 0.51,
      "learning_rate": 1e-05,
      "loss": 1.0174,
      "step": 357
    },
    {
      "epoch": 0.51,
      "learning_rate": 1e-05,
      "loss": 1.091,
      "step": 358
    },
    {
      "epoch": 0.51,
      "learning_rate": 1e-05,
      "loss": 0.9832,
      "step": 359
    },
    {
      "epoch": 0.52,
      "learning_rate": 1e-05,
      "loss": 1.115,
      "step": 360
    },
    {
      "epoch": 0.52,
      "learning_rate": 1e-05,
      "loss": 1.0466,
      "step": 361
    },
    {
      "epoch": 0.52,
      "learning_rate": 1e-05,
      "loss": 0.8839,
      "step": 362
    },
    {
      "epoch": 0.52,
      "learning_rate": 1e-05,
      "loss": 0.9968,
      "step": 363
    },
    {
      "epoch": 0.52,
      "learning_rate": 1e-05,
      "loss": 1.001,
      "step": 364
    },
    {
      "epoch": 0.52,
      "learning_rate": 1e-05,
      "loss": 1.0099,
      "step": 365
    },
    {
      "epoch": 0.52,
      "learning_rate": 1e-05,
      "loss": 1.0139,
      "step": 366
    },
    {
      "epoch": 0.53,
      "learning_rate": 1e-05,
      "loss": 0.9897,
      "step": 367
    },
    {
      "epoch": 0.53,
      "learning_rate": 1e-05,
      "loss": 0.9666,
      "step": 368
    },
    {
      "epoch": 0.53,
      "learning_rate": 1e-05,
      "loss": 1.0148,
      "step": 369
    },
    {
      "epoch": 0.53,
      "learning_rate": 1e-05,
      "loss": 1.1277,
      "step": 370
    },
    {
      "epoch": 0.53,
      "learning_rate": 1e-05,
      "loss": 1.0721,
      "step": 371
    },
    {
      "epoch": 0.53,
      "learning_rate": 1e-05,
      "loss": 1.0291,
      "step": 372
    },
    {
      "epoch": 0.53,
      "learning_rate": 1e-05,
      "loss": 1.0513,
      "step": 373
    },
    {
      "epoch": 0.54,
      "learning_rate": 1e-05,
      "loss": 1.012,
      "step": 374
    },
    {
      "epoch": 0.54,
      "learning_rate": 1e-05,
      "loss": 1.0807,
      "step": 375
    },
    {
      "epoch": 0.54,
      "learning_rate": 1e-05,
      "loss": 1.1254,
      "step": 376
    },
    {
      "epoch": 0.54,
      "learning_rate": 1e-05,
      "loss": 0.9723,
      "step": 377
    },
    {
      "epoch": 0.54,
      "learning_rate": 1e-05,
      "loss": 1.1322,
      "step": 378
    },
    {
      "epoch": 0.54,
      "learning_rate": 1e-05,
      "loss": 1.1148,
      "step": 379
    },
    {
      "epoch": 0.54,
      "learning_rate": 1e-05,
      "loss": 0.9667,
      "step": 380
    },
    {
      "epoch": 0.55,
      "learning_rate": 1e-05,
      "loss": 1.0855,
      "step": 381
    },
    {
      "epoch": 0.55,
      "learning_rate": 1e-05,
      "loss": 1.1321,
      "step": 382
    },
    {
      "epoch": 0.55,
      "learning_rate": 1e-05,
      "loss": 0.9415,
      "step": 383
    },
    {
      "epoch": 0.55,
      "learning_rate": 1e-05,
      "loss": 1.0164,
      "step": 384
    },
    {
      "epoch": 0.55,
      "learning_rate": 1e-05,
      "loss": 0.9239,
      "step": 385
    },
    {
      "epoch": 0.55,
      "learning_rate": 1e-05,
      "loss": 1.0676,
      "step": 386
    },
    {
      "epoch": 0.56,
      "learning_rate": 1e-05,
      "loss": 1.0899,
      "step": 387
    },
    {
      "epoch": 0.56,
      "learning_rate": 1e-05,
      "loss": 0.9679,
      "step": 388
    },
    {
      "epoch": 0.56,
      "learning_rate": 1e-05,
      "loss": 0.9323,
      "step": 389
    },
    {
      "epoch": 0.56,
      "learning_rate": 1e-05,
      "loss": 1.0108,
      "step": 390
    },
    {
      "epoch": 0.56,
      "learning_rate": 1e-05,
      "loss": 1.0285,
      "step": 391
    },
    {
      "epoch": 0.56,
      "learning_rate": 1e-05,
      "loss": 1.0663,
      "step": 392
    },
    {
      "epoch": 0.56,
      "learning_rate": 1e-05,
      "loss": 0.9585,
      "step": 393
    },
    {
      "epoch": 0.57,
      "learning_rate": 1e-05,
      "loss": 1.0131,
      "step": 394
    },
    {
      "epoch": 0.57,
      "learning_rate": 1e-05,
      "loss": 0.9502,
      "step": 395
    },
    {
      "epoch": 0.57,
      "learning_rate": 1e-05,
      "loss": 1.047,
      "step": 396
    },
    {
      "epoch": 0.57,
      "learning_rate": 1e-05,
      "loss": 0.9926,
      "step": 397
    },
    {
      "epoch": 0.57,
      "learning_rate": 1e-05,
      "loss": 1.1131,
      "step": 398
    },
    {
      "epoch": 0.57,
      "learning_rate": 1e-05,
      "loss": 1.1753,
      "step": 399
    },
    {
      "epoch": 0.57,
      "learning_rate": 1e-05,
      "loss": 0.9458,
      "step": 400
    },
    {
      "epoch": 0.58,
      "learning_rate": 1e-05,
      "loss": 1.0676,
      "step": 401
    },
    {
      "epoch": 0.58,
      "learning_rate": 1e-05,
      "loss": 0.972,
      "step": 402
    },
    {
      "epoch": 0.58,
      "learning_rate": 1e-05,
      "loss": 1.0739,
      "step": 403
    },
    {
      "epoch": 0.58,
      "learning_rate": 1e-05,
      "loss": 1.0843,
      "step": 404
    },
    {
      "epoch": 0.58,
      "learning_rate": 1e-05,
      "loss": 1.111,
      "step": 405
    },
    {
      "epoch": 0.58,
      "learning_rate": 1e-05,
      "loss": 1.1299,
      "step": 406
    },
    {
      "epoch": 0.58,
      "learning_rate": 1e-05,
      "loss": 0.9199,
      "step": 407
    },
    {
      "epoch": 0.59,
      "learning_rate": 1e-05,
      "loss": 1.0964,
      "step": 408
    },
    {
      "epoch": 0.59,
      "learning_rate": 1e-05,
      "loss": 0.9897,
      "step": 409
    },
    {
      "epoch": 0.59,
      "learning_rate": 1e-05,
      "loss": 0.9749,
      "step": 410
    },
    {
      "epoch": 0.59,
      "learning_rate": 1e-05,
      "loss": 1.0384,
      "step": 411
    },
    {
      "epoch": 0.59,
      "learning_rate": 1e-05,
      "loss": 0.9811,
      "step": 412
    },
    {
      "epoch": 0.59,
      "learning_rate": 1e-05,
      "loss": 0.9527,
      "step": 413
    },
    {
      "epoch": 0.59,
      "learning_rate": 1e-05,
      "loss": 1.0403,
      "step": 414
    },
    {
      "epoch": 0.6,
      "learning_rate": 1e-05,
      "loss": 1.0632,
      "step": 415
    },
    {
      "epoch": 0.6,
      "learning_rate": 1e-05,
      "loss": 0.9966,
      "step": 416
    },
    {
      "epoch": 0.6,
      "learning_rate": 1e-05,
      "loss": 1.0999,
      "step": 417
    },
    {
      "epoch": 0.6,
      "learning_rate": 1e-05,
      "loss": 1.1067,
      "step": 418
    },
    {
      "epoch": 0.6,
      "learning_rate": 1e-05,
      "loss": 1.2804,
      "step": 419
    },
    {
      "epoch": 0.6,
      "learning_rate": 1e-05,
      "loss": 1.0324,
      "step": 420
    },
    {
      "epoch": 0.6,
      "learning_rate": 1e-05,
      "loss": 0.9833,
      "step": 421
    },
    {
      "epoch": 0.61,
      "learning_rate": 1e-05,
      "loss": 1.1758,
      "step": 422
    },
    {
      "epoch": 0.61,
      "learning_rate": 1e-05,
      "loss": 1.0586,
      "step": 423
    },
    {
      "epoch": 0.61,
      "learning_rate": 1e-05,
      "loss": 0.9869,
      "step": 424
    },
    {
      "epoch": 0.61,
      "learning_rate": 1e-05,
      "loss": 0.9466,
      "step": 425
    },
    {
      "epoch": 0.61,
      "learning_rate": 1e-05,
      "loss": 1.0072,
      "step": 426
    },
    {
      "epoch": 0.61,
      "learning_rate": 1e-05,
      "loss": 1.039,
      "step": 427
    },
    {
      "epoch": 0.61,
      "learning_rate": 1e-05,
      "loss": 1.0721,
      "step": 428
    },
    {
      "epoch": 0.62,
      "learning_rate": 1e-05,
      "loss": 1.096,
      "step": 429
    },
    {
      "epoch": 0.62,
      "learning_rate": 1e-05,
      "loss": 0.9832,
      "step": 430
    },
    {
      "epoch": 0.62,
      "learning_rate": 1e-05,
      "loss": 1.0388,
      "step": 431
    },
    {
      "epoch": 0.62,
      "learning_rate": 1e-05,
      "loss": 1.1221,
      "step": 432
    },
    {
      "epoch": 0.62,
      "learning_rate": 1e-05,
      "loss": 0.9865,
      "step": 433
    },
    {
      "epoch": 0.62,
      "learning_rate": 1e-05,
      "loss": 0.9306,
      "step": 434
    },
    {
      "epoch": 0.62,
      "learning_rate": 1e-05,
      "loss": 1.0569,
      "step": 435
    },
    {
      "epoch": 0.63,
      "learning_rate": 1e-05,
      "loss": 0.9654,
      "step": 436
    },
    {
      "epoch": 0.63,
      "learning_rate": 1e-05,
      "loss": 1.0102,
      "step": 437
    },
    {
      "epoch": 0.63,
      "learning_rate": 1e-05,
      "loss": 0.9594,
      "step": 438
    },
    {
      "epoch": 0.63,
      "learning_rate": 1e-05,
      "loss": 1.1176,
      "step": 439
    },
    {
      "epoch": 0.63,
      "learning_rate": 1e-05,
      "loss": 1.0583,
      "step": 440
    },
    {
      "epoch": 0.63,
      "learning_rate": 1e-05,
      "loss": 1.1587,
      "step": 441
    },
    {
      "epoch": 0.63,
      "learning_rate": 1e-05,
      "loss": 1.1152,
      "step": 442
    },
    {
      "epoch": 0.64,
      "learning_rate": 1e-05,
      "loss": 1.1824,
      "step": 443
    },
    {
      "epoch": 0.64,
      "learning_rate": 1e-05,
      "loss": 1.0209,
      "step": 444
    },
    {
      "epoch": 0.64,
      "learning_rate": 1e-05,
      "loss": 1.0175,
      "step": 445
    },
    {
      "epoch": 0.64,
      "learning_rate": 1e-05,
      "loss": 1.2202,
      "step": 446
    },
    {
      "epoch": 0.64,
      "learning_rate": 1e-05,
      "loss": 1.0731,
      "step": 447
    },
    {
      "epoch": 0.64,
      "learning_rate": 1e-05,
      "loss": 1.0048,
      "step": 448
    },
    {
      "epoch": 0.64,
      "learning_rate": 1e-05,
      "loss": 1.1069,
      "step": 449
    },
    {
      "epoch": 0.65,
      "learning_rate": 1e-05,
      "loss": 1.1423,
      "step": 450
    },
    {
      "epoch": 0.65,
      "learning_rate": 1e-05,
      "loss": 1.0876,
      "step": 451
    },
    {
      "epoch": 0.65,
      "learning_rate": 1e-05,
      "loss": 1.0031,
      "step": 452
    },
    {
      "epoch": 0.65,
      "learning_rate": 1e-05,
      "loss": 0.9925,
      "step": 453
    },
    {
      "epoch": 0.65,
      "learning_rate": 1e-05,
      "loss": 1.0527,
      "step": 454
    },
    {
      "epoch": 0.65,
      "learning_rate": 1e-05,
      "loss": 1.1383,
      "step": 455
    },
    {
      "epoch": 0.65,
      "learning_rate": 1e-05,
      "loss": 1.075,
      "step": 456
    },
    {
      "epoch": 0.66,
      "learning_rate": 1e-05,
      "loss": 1.1538,
      "step": 457
    },
    {
      "epoch": 0.66,
      "learning_rate": 1e-05,
      "loss": 1.1247,
      "step": 458
    },
    {
      "epoch": 0.66,
      "learning_rate": 1e-05,
      "loss": 0.8969,
      "step": 459
    },
    {
      "epoch": 0.66,
      "learning_rate": 1e-05,
      "loss": 1.0402,
      "step": 460
    },
    {
      "epoch": 0.66,
      "learning_rate": 1e-05,
      "loss": 1.0698,
      "step": 461
    },
    {
      "epoch": 0.66,
      "learning_rate": 1e-05,
      "loss": 1.1228,
      "step": 462
    },
    {
      "epoch": 0.66,
      "learning_rate": 1e-05,
      "loss": 0.9226,
      "step": 463
    },
    {
      "epoch": 0.67,
      "learning_rate": 1e-05,
      "loss": 0.9402,
      "step": 464
    },
    {
      "epoch": 0.67,
      "learning_rate": 1e-05,
      "loss": 0.9455,
      "step": 465
    },
    {
      "epoch": 0.67,
      "learning_rate": 1e-05,
      "loss": 1.0964,
      "step": 466
    },
    {
      "epoch": 0.67,
      "learning_rate": 1e-05,
      "loss": 1.1479,
      "step": 467
    },
    {
      "epoch": 0.67,
      "learning_rate": 1e-05,
      "loss": 0.9546,
      "step": 468
    },
    {
      "epoch": 0.67,
      "learning_rate": 1e-05,
      "loss": 1.0024,
      "step": 469
    },
    {
      "epoch": 0.67,
      "learning_rate": 1e-05,
      "loss": 1.0385,
      "step": 470
    },
    {
      "epoch": 0.68,
      "learning_rate": 1e-05,
      "loss": 1.1487,
      "step": 471
    },
    {
      "epoch": 0.68,
      "learning_rate": 1e-05,
      "loss": 1.1335,
      "step": 472
    },
    {
      "epoch": 0.68,
      "learning_rate": 1e-05,
      "loss": 0.9505,
      "step": 473
    },
    {
      "epoch": 0.68,
      "learning_rate": 1e-05,
      "loss": 1.0131,
      "step": 474
    },
    {
      "epoch": 0.68,
      "learning_rate": 1e-05,
      "loss": 1.0207,
      "step": 475
    },
    {
      "epoch": 0.68,
      "learning_rate": 1e-05,
      "loss": 1.0146,
      "step": 476
    },
    {
      "epoch": 0.68,
      "learning_rate": 1e-05,
      "loss": 1.1573,
      "step": 477
    },
    {
      "epoch": 0.69,
      "learning_rate": 1e-05,
      "loss": 0.8931,
      "step": 478
    },
    {
      "epoch": 0.69,
      "learning_rate": 1e-05,
      "loss": 0.9418,
      "step": 479
    },
    {
      "epoch": 0.69,
      "learning_rate": 1e-05,
      "loss": 1.7308,
      "step": 480
    },
    {
      "epoch": 0.69,
      "learning_rate": 1e-05,
      "loss": 1.0176,
      "step": 481
    },
    {
      "epoch": 0.69,
      "learning_rate": 1e-05,
      "loss": 1.0483,
      "step": 482
    },
    {
      "epoch": 0.69,
      "learning_rate": 1e-05,
      "loss": 1.0021,
      "step": 483
    },
    {
      "epoch": 0.69,
      "learning_rate": 1e-05,
      "loss": 1.085,
      "step": 484
    },
    {
      "epoch": 0.7,
      "learning_rate": 1e-05,
      "loss": 1.0358,
      "step": 485
    },
    {
      "epoch": 0.7,
      "learning_rate": 1e-05,
      "loss": 0.9895,
      "step": 486
    },
    {
      "epoch": 0.7,
      "learning_rate": 1e-05,
      "loss": 1.0381,
      "step": 487
    },
    {
      "epoch": 0.7,
      "learning_rate": 1e-05,
      "loss": 0.9967,
      "step": 488
    },
    {
      "epoch": 0.7,
      "learning_rate": 1e-05,
      "loss": 1.0472,
      "step": 489
    },
    {
      "epoch": 0.7,
      "learning_rate": 1e-05,
      "loss": 1.0044,
      "step": 490
    },
    {
      "epoch": 0.7,
      "learning_rate": 1e-05,
      "loss": 1.0401,
      "step": 491
    },
    {
      "epoch": 0.71,
      "learning_rate": 1e-05,
      "loss": 0.9565,
      "step": 492
    },
    {
      "epoch": 0.71,
      "learning_rate": 1e-05,
      "loss": 1.2636,
      "step": 493
    },
    {
      "epoch": 0.71,
      "learning_rate": 1e-05,
      "loss": 1.0165,
      "step": 494
    },
    {
      "epoch": 0.71,
      "learning_rate": 1e-05,
      "loss": 1.1178,
      "step": 495
    },
    {
      "epoch": 0.71,
      "learning_rate": 1e-05,
      "loss": 0.958,
      "step": 496
    },
    {
      "epoch": 0.71,
      "learning_rate": 1e-05,
      "loss": 1.2203,
      "step": 497
    },
    {
      "epoch": 0.71,
      "learning_rate": 1e-05,
      "loss": 0.8773,
      "step": 498
    },
    {
      "epoch": 0.72,
      "learning_rate": 1e-05,
      "loss": 0.9468,
      "step": 499
    },
    {
      "epoch": 0.72,
      "learning_rate": 1e-05,
      "loss": 0.9328,
      "step": 500
    },
    {
      "epoch": 0.72,
      "learning_rate": 1e-05,
      "loss": 0.9628,
      "step": 501
    },
    {
      "epoch": 0.72,
      "learning_rate": 1e-05,
      "loss": 0.9855,
      "step": 502
    },
    {
      "epoch": 0.72,
      "learning_rate": 1e-05,
      "loss": 1.0372,
      "step": 503
    },
    {
      "epoch": 0.72,
      "learning_rate": 1e-05,
      "loss": 1.1009,
      "step": 504
    },
    {
      "epoch": 0.72,
      "learning_rate": 1e-05,
      "loss": 1.0288,
      "step": 505
    },
    {
      "epoch": 0.73,
      "learning_rate": 1e-05,
      "loss": 0.9926,
      "step": 506
    },
    {
      "epoch": 0.73,
      "learning_rate": 1e-05,
      "loss": 1.053,
      "step": 507
    },
    {
      "epoch": 0.73,
      "learning_rate": 1e-05,
      "loss": 1.0558,
      "step": 508
    },
    {
      "epoch": 0.73,
      "learning_rate": 1e-05,
      "loss": 1.0034,
      "step": 509
    },
    {
      "epoch": 0.73,
      "learning_rate": 1e-05,
      "loss": 1.1113,
      "step": 510
    },
    {
      "epoch": 0.73,
      "learning_rate": 1e-05,
      "loss": 1.0469,
      "step": 511
    },
    {
      "epoch": 0.73,
      "learning_rate": 1e-05,
      "loss": 1.1266,
      "step": 512
    },
    {
      "epoch": 0.74,
      "learning_rate": 1e-05,
      "loss": 1.19,
      "step": 513
    },
    {
      "epoch": 0.74,
      "learning_rate": 1e-05,
      "loss": 0.9687,
      "step": 514
    },
    {
      "epoch": 0.74,
      "learning_rate": 1e-05,
      "loss": 1.0508,
      "step": 515
    },
    {
      "epoch": 0.74,
      "learning_rate": 1e-05,
      "loss": 1.0228,
      "step": 516
    },
    {
      "epoch": 0.74,
      "learning_rate": 1e-05,
      "loss": 0.8856,
      "step": 517
    },
    {
      "epoch": 0.74,
      "learning_rate": 1e-05,
      "loss": 1.1215,
      "step": 518
    },
    {
      "epoch": 0.74,
      "learning_rate": 1e-05,
      "loss": 1.003,
      "step": 519
    },
    {
      "epoch": 0.75,
      "learning_rate": 1e-05,
      "loss": 0.9588,
      "step": 520
    },
    {
      "epoch": 0.75,
      "learning_rate": 1e-05,
      "loss": 0.9806,
      "step": 521
    },
    {
      "epoch": 0.75,
      "learning_rate": 1e-05,
      "loss": 1.0773,
      "step": 522
    },
    {
      "epoch": 0.75,
      "learning_rate": 1e-05,
      "loss": 1.1008,
      "step": 523
    },
    {
      "epoch": 0.75,
      "learning_rate": 1e-05,
      "loss": 1.0028,
      "step": 524
    },
    {
      "epoch": 0.75,
      "learning_rate": 1e-05,
      "loss": 0.9927,
      "step": 525
    },
    {
      "epoch": 0.75,
      "learning_rate": 1e-05,
      "loss": 1.0067,
      "step": 526
    },
    {
      "epoch": 0.76,
      "learning_rate": 1e-05,
      "loss": 0.9375,
      "step": 527
    },
    {
      "epoch": 0.76,
      "learning_rate": 1e-05,
      "loss": 1.1179,
      "step": 528
    },
    {
      "epoch": 0.76,
      "learning_rate": 1e-05,
      "loss": 1.1061,
      "step": 529
    },
    {
      "epoch": 0.76,
      "learning_rate": 1e-05,
      "loss": 1.0092,
      "step": 530
    },
    {
      "epoch": 0.76,
      "learning_rate": 1e-05,
      "loss": 0.9984,
      "step": 531
    },
    {
      "epoch": 0.76,
      "learning_rate": 1e-05,
      "loss": 0.9813,
      "step": 532
    },
    {
      "epoch": 0.76,
      "learning_rate": 1e-05,
      "loss": 1.0271,
      "step": 533
    },
    {
      "epoch": 0.77,
      "learning_rate": 1e-05,
      "loss": 1.0579,
      "step": 534
    },
    {
      "epoch": 0.77,
      "learning_rate": 1e-05,
      "loss": 1.2079,
      "step": 535
    },
    {
      "epoch": 0.77,
      "learning_rate": 1e-05,
      "loss": 1.0207,
      "step": 536
    },
    {
      "epoch": 0.77,
      "learning_rate": 1e-05,
      "loss": 0.935,
      "step": 537
    },
    {
      "epoch": 0.77,
      "learning_rate": 1e-05,
      "loss": 0.9786,
      "step": 538
    },
    {
      "epoch": 0.77,
      "learning_rate": 1e-05,
      "loss": 0.9921,
      "step": 539
    },
    {
      "epoch": 0.77,
      "learning_rate": 1e-05,
      "loss": 0.9515,
      "step": 540
    },
    {
      "epoch": 0.78,
      "learning_rate": 1e-05,
      "loss": 0.9562,
      "step": 541
    },
    {
      "epoch": 0.78,
      "learning_rate": 1e-05,
      "loss": 0.9814,
      "step": 542
    },
    {
      "epoch": 0.78,
      "learning_rate": 1e-05,
      "loss": 1.063,
      "step": 543
    },
    {
      "epoch": 0.78,
      "learning_rate": 1e-05,
      "loss": 1.0257,
      "step": 544
    },
    {
      "epoch": 0.78,
      "learning_rate": 1e-05,
      "loss": 0.9897,
      "step": 545
    },
    {
      "epoch": 0.78,
      "learning_rate": 1e-05,
      "loss": 0.9595,
      "step": 546
    },
    {
      "epoch": 0.78,
      "learning_rate": 1e-05,
      "loss": 0.9941,
      "step": 547
    },
    {
      "epoch": 0.79,
      "learning_rate": 1e-05,
      "loss": 1.007,
      "step": 548
    },
    {
      "epoch": 0.79,
      "learning_rate": 1e-05,
      "loss": 1.0209,
      "step": 549
    },
    {
      "epoch": 0.79,
      "learning_rate": 1e-05,
      "loss": 0.987,
      "step": 550
    },
    {
      "epoch": 0.79,
      "learning_rate": 1e-05,
      "loss": 1.0833,
      "step": 551
    },
    {
      "epoch": 0.79,
      "learning_rate": 1e-05,
      "loss": 1.1884,
      "step": 552
    },
    {
      "epoch": 0.79,
      "learning_rate": 1e-05,
      "loss": 1.1615,
      "step": 553
    },
    {
      "epoch": 0.79,
      "learning_rate": 1e-05,
      "loss": 0.9579,
      "step": 554
    },
    {
      "epoch": 0.8,
      "learning_rate": 1e-05,
      "loss": 0.9642,
      "step": 555
    },
    {
      "epoch": 0.8,
      "learning_rate": 1e-05,
      "loss": 0.9798,
      "step": 556
    },
    {
      "epoch": 0.8,
      "learning_rate": 1e-05,
      "loss": 0.9784,
      "step": 557
    },
    {
      "epoch": 0.8,
      "learning_rate": 1e-05,
      "loss": 1.1169,
      "step": 558
    },
    {
      "epoch": 0.8,
      "learning_rate": 1e-05,
      "loss": 1.0233,
      "step": 559
    },
    {
      "epoch": 0.8,
      "learning_rate": 1e-05,
      "loss": 1.01,
      "step": 560
    },
    {
      "epoch": 0.8,
      "learning_rate": 1e-05,
      "loss": 1.0844,
      "step": 561
    },
    {
      "epoch": 0.81,
      "learning_rate": 1e-05,
      "loss": 1.0328,
      "step": 562
    },
    {
      "epoch": 0.81,
      "learning_rate": 1e-05,
      "loss": 0.9969,
      "step": 563
    },
    {
      "epoch": 0.81,
      "learning_rate": 1e-05,
      "loss": 1.1353,
      "step": 564
    },
    {
      "epoch": 0.81,
      "learning_rate": 1e-05,
      "loss": 0.9913,
      "step": 565
    },
    {
      "epoch": 0.81,
      "learning_rate": 1e-05,
      "loss": 0.9789,
      "step": 566
    },
    {
      "epoch": 0.81,
      "learning_rate": 1e-05,
      "loss": 0.9884,
      "step": 567
    },
    {
      "epoch": 0.81,
      "learning_rate": 1e-05,
      "loss": 0.9269,
      "step": 568
    },
    {
      "epoch": 0.82,
      "learning_rate": 1e-05,
      "loss": 0.9541,
      "step": 569
    },
    {
      "epoch": 0.82,
      "learning_rate": 1e-05,
      "loss": 1.0094,
      "step": 570
    },
    {
      "epoch": 0.82,
      "learning_rate": 1e-05,
      "loss": 0.9262,
      "step": 571
    },
    {
      "epoch": 0.82,
      "learning_rate": 1e-05,
      "loss": 1.0006,
      "step": 572
    },
    {
      "epoch": 0.82,
      "learning_rate": 1e-05,
      "loss": 1.0208,
      "step": 573
    },
    {
      "epoch": 0.82,
      "learning_rate": 1e-05,
      "loss": 0.9545,
      "step": 574
    },
    {
      "epoch": 0.82,
      "learning_rate": 1e-05,
      "loss": 1.0529,
      "step": 575
    },
    {
      "epoch": 0.83,
      "learning_rate": 1e-05,
      "loss": 1.0522,
      "step": 576
    },
    {
      "epoch": 0.83,
      "learning_rate": 1e-05,
      "loss": 0.994,
      "step": 577
    },
    {
      "epoch": 0.83,
      "learning_rate": 1e-05,
      "loss": 1.0601,
      "step": 578
    },
    {
      "epoch": 0.83,
      "learning_rate": 1e-05,
      "loss": 1.0488,
      "step": 579
    },
    {
      "epoch": 0.83,
      "learning_rate": 1e-05,
      "loss": 1.1094,
      "step": 580
    },
    {
      "epoch": 0.83,
      "learning_rate": 1e-05,
      "loss": 0.9792,
      "step": 581
    },
    {
      "epoch": 0.83,
      "learning_rate": 1e-05,
      "loss": 0.9355,
      "step": 582
    },
    {
      "epoch": 0.84,
      "learning_rate": 1e-05,
      "loss": 0.9976,
      "step": 583
    },
    {
      "epoch": 0.84,
      "learning_rate": 1e-05,
      "loss": 1.0626,
      "step": 584
    },
    {
      "epoch": 0.84,
      "learning_rate": 1e-05,
      "loss": 1.0379,
      "step": 585
    },
    {
      "epoch": 0.84,
      "learning_rate": 1e-05,
      "loss": 0.9992,
      "step": 586
    },
    {
      "epoch": 0.84,
      "learning_rate": 1e-05,
      "loss": 1.0146,
      "step": 587
    },
    {
      "epoch": 0.84,
      "learning_rate": 1e-05,
      "loss": 1.0091,
      "step": 588
    },
    {
      "epoch": 0.84,
      "learning_rate": 1e-05,
      "loss": 0.9756,
      "step": 589
    },
    {
      "epoch": 0.85,
      "learning_rate": 1e-05,
      "loss": 1.0304,
      "step": 590
    },
    {
      "epoch": 0.85,
      "learning_rate": 1e-05,
      "loss": 0.9871,
      "step": 591
    },
    {
      "epoch": 0.85,
      "learning_rate": 1e-05,
      "loss": 0.9624,
      "step": 592
    },
    {
      "epoch": 0.85,
      "learning_rate": 1e-05,
      "loss": 1.0457,
      "step": 593
    },
    {
      "epoch": 0.85,
      "learning_rate": 1e-05,
      "loss": 1.0368,
      "step": 594
    },
    {
      "epoch": 0.85,
      "learning_rate": 1e-05,
      "loss": 1.021,
      "step": 595
    },
    {
      "epoch": 0.85,
      "learning_rate": 1e-05,
      "loss": 0.9531,
      "step": 596
    },
    {
      "epoch": 0.86,
      "learning_rate": 1e-05,
      "loss": 1.0019,
      "step": 597
    },
    {
      "epoch": 0.86,
      "learning_rate": 1e-05,
      "loss": 1.043,
      "step": 598
    },
    {
      "epoch": 0.86,
      "learning_rate": 1e-05,
      "loss": 0.9439,
      "step": 599
    },
    {
      "epoch": 0.86,
      "learning_rate": 1e-05,
      "loss": 0.9815,
      "step": 600
    },
    {
      "epoch": 0.86,
      "learning_rate": 1e-05,
      "loss": 1.0169,
      "step": 601
    },
    {
      "epoch": 0.86,
      "learning_rate": 1e-05,
      "loss": 1.1024,
      "step": 602
    },
    {
      "epoch": 0.86,
      "learning_rate": 1e-05,
      "loss": 1.0489,
      "step": 603
    },
    {
      "epoch": 0.87,
      "learning_rate": 1e-05,
      "loss": 1.056,
      "step": 604
    },
    {
      "epoch": 0.87,
      "learning_rate": 1e-05,
      "loss": 1.0706,
      "step": 605
    },
    {
      "epoch": 0.87,
      "learning_rate": 1e-05,
      "loss": 1.0587,
      "step": 606
    },
    {
      "epoch": 0.87,
      "learning_rate": 1e-05,
      "loss": 1.1469,
      "step": 607
    },
    {
      "epoch": 0.87,
      "learning_rate": 1e-05,
      "loss": 1.0443,
      "step": 608
    },
    {
      "epoch": 0.87,
      "learning_rate": 1e-05,
      "loss": 1.0336,
      "step": 609
    },
    {
      "epoch": 0.87,
      "learning_rate": 1e-05,
      "loss": 1.0307,
      "step": 610
    },
    {
      "epoch": 0.88,
      "learning_rate": 1e-05,
      "loss": 0.9896,
      "step": 611
    },
    {
      "epoch": 0.88,
      "learning_rate": 1e-05,
      "loss": 0.9988,
      "step": 612
    },
    {
      "epoch": 0.88,
      "learning_rate": 1e-05,
      "loss": 1.0166,
      "step": 613
    },
    {
      "epoch": 0.88,
      "learning_rate": 1e-05,
      "loss": 0.9082,
      "step": 614
    },
    {
      "epoch": 0.88,
      "learning_rate": 1e-05,
      "loss": 1.0225,
      "step": 615
    },
    {
      "epoch": 0.88,
      "learning_rate": 1e-05,
      "loss": 0.8942,
      "step": 616
    },
    {
      "epoch": 0.88,
      "learning_rate": 1e-05,
      "loss": 1.0004,
      "step": 617
    },
    {
      "epoch": 0.89,
      "learning_rate": 1e-05,
      "loss": 1.2904,
      "step": 618
    },
    {
      "epoch": 0.89,
      "learning_rate": 1e-05,
      "loss": 0.9123,
      "step": 619
    },
    {
      "epoch": 0.89,
      "learning_rate": 1e-05,
      "loss": 1.0391,
      "step": 620
    },
    {
      "epoch": 0.89,
      "learning_rate": 1e-05,
      "loss": 0.9553,
      "step": 621
    },
    {
      "epoch": 0.89,
      "learning_rate": 1e-05,
      "loss": 1.0575,
      "step": 622
    },
    {
      "epoch": 0.89,
      "learning_rate": 1e-05,
      "loss": 1.0035,
      "step": 623
    },
    {
      "epoch": 0.89,
      "learning_rate": 1e-05,
      "loss": 1.0506,
      "step": 624
    },
    {
      "epoch": 0.9,
      "learning_rate": 1e-05,
      "loss": 1.0033,
      "step": 625
    },
    {
      "epoch": 0.9,
      "learning_rate": 1e-05,
      "loss": 1.0883,
      "step": 626
    },
    {
      "epoch": 0.9,
      "learning_rate": 1e-05,
      "loss": 0.9317,
      "step": 627
    },
    {
      "epoch": 0.9,
      "learning_rate": 1e-05,
      "loss": 1.1029,
      "step": 628
    },
    {
      "epoch": 0.9,
      "learning_rate": 1e-05,
      "loss": 0.8631,
      "step": 629
    },
    {
      "epoch": 0.9,
      "learning_rate": 1e-05,
      "loss": 1.0515,
      "step": 630
    },
    {
      "epoch": 0.9,
      "learning_rate": 1e-05,
      "loss": 0.8643,
      "step": 631
    },
    {
      "epoch": 0.91,
      "learning_rate": 1e-05,
      "loss": 0.9945,
      "step": 632
    },
    {
      "epoch": 0.91,
      "learning_rate": 1e-05,
      "loss": 1.0214,
      "step": 633
    },
    {
      "epoch": 0.91,
      "learning_rate": 1e-05,
      "loss": 1.0223,
      "step": 634
    },
    {
      "epoch": 0.91,
      "learning_rate": 1e-05,
      "loss": 1.0406,
      "step": 635
    },
    {
      "epoch": 0.91,
      "learning_rate": 1e-05,
      "loss": 0.8677,
      "step": 636
    },
    {
      "epoch": 0.91,
      "learning_rate": 1e-05,
      "loss": 1.0497,
      "step": 637
    },
    {
      "epoch": 0.92,
      "learning_rate": 1e-05,
      "loss": 1.0703,
      "step": 638
    },
    {
      "epoch": 0.92,
      "learning_rate": 1e-05,
      "loss": 1.0756,
      "step": 639
    },
    {
      "epoch": 0.92,
      "learning_rate": 1e-05,
      "loss": 1.3689,
      "step": 640
    },
    {
      "epoch": 0.92,
      "learning_rate": 1e-05,
      "loss": 1.0132,
      "step": 641
    },
    {
      "epoch": 0.92,
      "learning_rate": 1e-05,
      "loss": 1.0111,
      "step": 642
    },
    {
      "epoch": 0.92,
      "learning_rate": 1e-05,
      "loss": 1.008,
      "step": 643
    },
    {
      "epoch": 0.92,
      "learning_rate": 1e-05,
      "loss": 1.0341,
      "step": 644
    },
    {
      "epoch": 0.93,
      "learning_rate": 1e-05,
      "loss": 1.0253,
      "step": 645
    },
    {
      "epoch": 0.93,
      "learning_rate": 1e-05,
      "loss": 0.9846,
      "step": 646
    },
    {
      "epoch": 0.93,
      "learning_rate": 1e-05,
      "loss": 1.0045,
      "step": 647
    },
    {
      "epoch": 0.93,
      "learning_rate": 1e-05,
      "loss": 0.9847,
      "step": 648
    },
    {
      "epoch": 0.93,
      "learning_rate": 1e-05,
      "loss": 1.0752,
      "step": 649
    },
    {
      "epoch": 0.93,
      "learning_rate": 1e-05,
      "loss": 0.9741,
      "step": 650
    },
    {
      "epoch": 0.93,
      "learning_rate": 1e-05,
      "loss": 1.0448,
      "step": 651
    },
    {
      "epoch": 0.94,
      "learning_rate": 1e-05,
      "loss": 1.0645,
      "step": 652
    },
    {
      "epoch": 0.94,
      "learning_rate": 1e-05,
      "loss": 0.9702,
      "step": 653
    },
    {
      "epoch": 0.94,
      "learning_rate": 1e-05,
      "loss": 1.186,
      "step": 654
    },
    {
      "epoch": 0.94,
      "learning_rate": 1e-05,
      "loss": 0.9269,
      "step": 655
    },
    {
      "epoch": 0.94,
      "learning_rate": 1e-05,
      "loss": 0.9912,
      "step": 656
    },
    {
      "epoch": 0.94,
      "learning_rate": 1e-05,
      "loss": 1.072,
      "step": 657
    },
    {
      "epoch": 0.94,
      "learning_rate": 1e-05,
      "loss": 0.9298,
      "step": 658
    },
    {
      "epoch": 0.95,
      "learning_rate": 1e-05,
      "loss": 1.1102,
      "step": 659
    },
    {
      "epoch": 0.95,
      "learning_rate": 1e-05,
      "loss": 0.9396,
      "step": 660
    },
    {
      "epoch": 0.95,
      "learning_rate": 1e-05,
      "loss": 0.9688,
      "step": 661
    },
    {
      "epoch": 0.95,
      "learning_rate": 1e-05,
      "loss": 1.0481,
      "step": 662
    },
    {
      "epoch": 0.95,
      "learning_rate": 1e-05,
      "loss": 0.9592,
      "step": 663
    },
    {
      "epoch": 0.95,
      "learning_rate": 1e-05,
      "loss": 1.1473,
      "step": 664
    },
    {
      "epoch": 0.95,
      "learning_rate": 1e-05,
      "loss": 1.06,
      "step": 665
    },
    {
      "epoch": 0.96,
      "learning_rate": 1e-05,
      "loss": 1.0454,
      "step": 666
    },
    {
      "epoch": 0.96,
      "learning_rate": 1e-05,
      "loss": 1.0591,
      "step": 667
    },
    {
      "epoch": 0.96,
      "learning_rate": 1e-05,
      "loss": 0.9713,
      "step": 668
    },
    {
      "epoch": 0.96,
      "learning_rate": 1e-05,
      "loss": 1.0062,
      "step": 669
    },
    {
      "epoch": 0.96,
      "learning_rate": 1e-05,
      "loss": 1.0873,
      "step": 670
    },
    {
      "epoch": 0.96,
      "learning_rate": 1e-05,
      "loss": 0.9474,
      "step": 671
    },
    {
      "epoch": 0.96,
      "learning_rate": 1e-05,
      "loss": 0.9709,
      "step": 672
    },
    {
      "epoch": 0.97,
      "learning_rate": 1e-05,
      "loss": 1.0629,
      "step": 673
    },
    {
      "epoch": 0.97,
      "learning_rate": 1e-05,
      "loss": 1.0446,
      "step": 674
    },
    {
      "epoch": 0.97,
      "learning_rate": 1e-05,
      "loss": 0.9441,
      "step": 675
    },
    {
      "epoch": 0.97,
      "learning_rate": 1e-05,
      "loss": 0.998,
      "step": 676
    },
    {
      "epoch": 0.97,
      "learning_rate": 1e-05,
      "loss": 1.0394,
      "step": 677
    },
    {
      "epoch": 0.97,
      "learning_rate": 1e-05,
      "loss": 1.1014,
      "step": 678
    },
    {
      "epoch": 0.97,
      "learning_rate": 1e-05,
      "loss": 1.1001,
      "step": 679
    },
    {
      "epoch": 0.98,
      "learning_rate": 1e-05,
      "loss": 1.0624,
      "step": 680
    },
    {
      "epoch": 0.98,
      "learning_rate": 1e-05,
      "loss": 1.0151,
      "step": 681
    },
    {
      "epoch": 0.98,
      "learning_rate": 1e-05,
      "loss": 1.0519,
      "step": 682
    },
    {
      "epoch": 0.98,
      "learning_rate": 1e-05,
      "loss": 1.0329,
      "step": 683
    },
    {
      "epoch": 0.98,
      "learning_rate": 1e-05,
      "loss": 1.13,
      "step": 684
    },
    {
      "epoch": 0.98,
      "learning_rate": 1e-05,
      "loss": 0.9441,
      "step": 685
    },
    {
      "epoch": 0.98,
      "learning_rate": 1e-05,
      "loss": 1.0608,
      "step": 686
    },
    {
      "epoch": 0.99,
      "learning_rate": 1e-05,
      "loss": 0.9612,
      "step": 687
    },
    {
      "epoch": 0.99,
      "learning_rate": 1e-05,
      "loss": 1.08,
      "step": 688
    },
    {
      "epoch": 0.99,
      "learning_rate": 1e-05,
      "loss": 1.0851,
      "step": 689
    },
    {
      "epoch": 0.99,
      "learning_rate": 1e-05,
      "loss": 1.0611,
      "step": 690
    },
    {
      "epoch": 0.99,
      "learning_rate": 1e-05,
      "loss": 1.0158,
      "step": 691
    },
    {
      "epoch": 0.99,
      "learning_rate": 1e-05,
      "loss": 1.0013,
      "step": 692
    },
    {
      "epoch": 0.99,
      "learning_rate": 1e-05,
      "loss": 1.1202,
      "step": 693
    },
    {
      "epoch": 1.0,
      "learning_rate": 1e-05,
      "loss": 1.247,
      "step": 694
    },
    {
      "epoch": 1.0,
      "learning_rate": 1e-05,
      "loss": 0.979,
      "step": 695
    },
    {
      "epoch": 1.0,
      "learning_rate": 1e-05,
      "loss": 1.0387,
      "step": 696
    },
    {
      "epoch": 1.0,
      "learning_rate": 1e-05,
      "loss": 1.0131,
      "step": 697
    },
    {
      "epoch": 1.0,
      "learning_rate": 1e-05,
      "loss": 1.2666,
      "step": 698
    },
    {
      "epoch": 1.0,
      "learning_rate": 1e-05,
      "loss": 0.9538,
      "step": 699
    },
    {
      "epoch": 1.0,
      "learning_rate": 1e-05,
      "loss": 1.065,
      "step": 700
    },
    {
      "epoch": 1.01,
      "learning_rate": 1e-05,
      "loss": 0.992,
      "step": 701
    },
    {
      "epoch": 1.01,
      "learning_rate": 1e-05,
      "loss": 1.055,
      "step": 702
    },
    {
      "epoch": 1.01,
      "learning_rate": 1e-05,
      "loss": 1.1001,
      "step": 703
    },
    {
      "epoch": 1.01,
      "learning_rate": 1e-05,
      "loss": 1.0113,
      "step": 704
    },
    {
      "epoch": 1.01,
      "learning_rate": 1e-05,
      "loss": 1.0397,
      "step": 705
    },
    {
      "epoch": 1.01,
      "learning_rate": 1e-05,
      "loss": 1.0019,
      "step": 706
    },
    {
      "epoch": 1.01,
      "learning_rate": 1e-05,
      "loss": 0.969,
      "step": 707
    },
    {
      "epoch": 1.02,
      "learning_rate": 1e-05,
      "loss": 0.9365,
      "step": 708
    },
    {
      "epoch": 1.02,
      "learning_rate": 1e-05,
      "loss": 0.9784,
      "step": 709
    },
    {
      "epoch": 1.02,
      "learning_rate": 1e-05,
      "loss": 0.9263,
      "step": 710
    }
  ],
  "max_steps": 69700,
  "num_train_epochs": 100,
  "total_flos": 5.110698589802496e+16,
  "trial_name": null,
  "trial_params": null
}