{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.9987389659520807,
  "global_step": 297,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "learning_rate": 0.0,
      "loss": 8.889,
      "step": 1
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0,
      "loss": 8.8304,
      "step": 2
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0,
      "loss": 8.8708,
      "step": 3
    },
    {
      "epoch": 0.01,
      "learning_rate": 6.000000000000001e-08,
      "loss": 8.8472,
      "step": 4
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.2000000000000002e-07,
      "loss": 8.7009,
      "step": 5
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.8e-07,
      "loss": 8.7831,
      "step": 6
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.4000000000000003e-07,
      "loss": 8.9204,
      "step": 7
    },
    {
      "epoch": 0.03,
      "learning_rate": 3.0000000000000004e-07,
      "loss": 8.7855,
      "step": 8
    },
    {
      "epoch": 0.03,
      "learning_rate": 3.0000000000000004e-07,
      "loss": 8.996,
      "step": 9
    },
    {
      "epoch": 0.03,
      "learning_rate": 3.6e-07,
      "loss": 8.7342,
      "step": 10
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.2e-07,
      "loss": 8.658,
      "step": 11
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.800000000000001e-07,
      "loss": 8.6911,
      "step": 12
    },
    {
      "epoch": 0.04,
      "learning_rate": 5.4e-07,
      "loss": 8.6838,
      "step": 13
    },
    {
      "epoch": 0.05,
      "learning_rate": 6.000000000000001e-07,
      "loss": 8.6734,
      "step": 14
    },
    {
      "epoch": 0.05,
      "learning_rate": 6.599999999999999e-07,
      "loss": 8.552,
      "step": 15
    },
    {
      "epoch": 0.05,
      "learning_rate": 7.2e-07,
      "loss": 8.3643,
      "step": 16
    },
    {
      "epoch": 0.06,
      "learning_rate": 7.799999999999999e-07,
      "loss": 8.3339,
      "step": 17
    },
    {
      "epoch": 0.06,
      "learning_rate": 8.4e-07,
      "loss": 8.3624,
      "step": 18
    },
    {
      "epoch": 0.06,
      "learning_rate": 9e-07,
      "loss": 8.2912,
      "step": 19
    },
    {
      "epoch": 0.07,
      "learning_rate": 9.600000000000001e-07,
      "loss": 8.2759,
      "step": 20
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.0200000000000002e-06,
      "loss": 8.1023,
      "step": 21
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.08e-06,
      "loss": 8.1226,
      "step": 22
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.14e-06,
      "loss": 7.9866,
      "step": 23
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.2000000000000002e-06,
      "loss": 8.1126,
      "step": 24
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.26e-06,
      "loss": 8.0722,
      "step": 25
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.3199999999999999e-06,
      "loss": 7.9731,
      "step": 26
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.38e-06,
      "loss": 7.8968,
      "step": 27
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.44e-06,
      "loss": 7.8195,
      "step": 28
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.5e-06,
      "loss": 7.7153,
      "step": 29
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.5599999999999999e-06,
      "loss": 7.7519,
      "step": 30
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.62e-06,
      "loss": 7.7105,
      "step": 31
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.68e-06,
      "loss": 7.7445,
      "step": 32
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.74e-06,
      "loss": 7.629,
      "step": 33
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.8e-06,
      "loss": 7.5415,
      "step": 34
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.86e-06,
      "loss": 7.5895,
      "step": 35
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.9200000000000003e-06,
      "loss": 7.5373,
      "step": 36
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.98e-06,
      "loss": 7.5932,
      "step": 37
    },
    {
      "epoch": 0.13,
      "learning_rate": 2.0400000000000004e-06,
      "loss": 7.5453,
      "step": 38
    },
    {
      "epoch": 0.13,
      "learning_rate": 2.1000000000000002e-06,
      "loss": 7.5879,
      "step": 39
    },
    {
      "epoch": 0.13,
      "learning_rate": 2.16e-06,
      "loss": 7.5368,
      "step": 40
    },
    {
      "epoch": 0.14,
      "learning_rate": 2.22e-06,
      "loss": 7.6058,
      "step": 41
    },
    {
      "epoch": 0.14,
      "learning_rate": 2.28e-06,
      "loss": 7.5286,
      "step": 42
    },
    {
      "epoch": 0.14,
      "learning_rate": 2.34e-06,
      "loss": 7.4283,
      "step": 43
    },
    {
      "epoch": 0.15,
      "learning_rate": 2.4000000000000003e-06,
      "loss": 7.6536,
      "step": 44
    },
    {
      "epoch": 0.15,
      "learning_rate": 2.46e-06,
      "loss": 7.5623,
      "step": 45
    },
    {
      "epoch": 0.15,
      "learning_rate": 2.52e-06,
      "loss": 7.5319,
      "step": 46
    },
    {
      "epoch": 0.16,
      "learning_rate": 2.58e-06,
      "loss": 7.6,
      "step": 47
    },
    {
      "epoch": 0.16,
      "learning_rate": 2.6399999999999997e-06,
      "loss": 7.5462,
      "step": 48
    },
    {
      "epoch": 0.16,
      "learning_rate": 2.7e-06,
      "loss": 7.5795,
      "step": 49
    },
    {
      "epoch": 0.17,
      "learning_rate": 2.76e-06,
      "loss": 7.5914,
      "step": 50
    },
    {
      "epoch": 0.17,
      "learning_rate": 2.82e-06,
      "loss": 7.3664,
      "step": 51
    },
    {
      "epoch": 0.17,
      "learning_rate": 2.88e-06,
      "loss": 7.3437,
      "step": 52
    },
    {
      "epoch": 0.18,
      "learning_rate": 2.9400000000000002e-06,
      "loss": 7.2641,
      "step": 53
    },
    {
      "epoch": 0.18,
      "learning_rate": 3e-06,
      "loss": 7.2696,
      "step": 54
    },
    {
      "epoch": 0.18,
      "learning_rate": 3.06e-06,
      "loss": 7.2381,
      "step": 55
    },
    {
      "epoch": 0.19,
      "learning_rate": 3.1199999999999998e-06,
      "loss": 7.2935,
      "step": 56
    },
    {
      "epoch": 0.19,
      "learning_rate": 3.18e-06,
      "loss": 7.3188,
      "step": 57
    },
    {
      "epoch": 0.2,
      "learning_rate": 3.24e-06,
      "loss": 7.3463,
      "step": 58
    },
    {
      "epoch": 0.2,
      "learning_rate": 3.3e-06,
      "loss": 7.2282,
      "step": 59
    },
    {
      "epoch": 0.2,
      "learning_rate": 3.36e-06,
      "loss": 7.2696,
      "step": 60
    },
    {
      "epoch": 0.21,
      "learning_rate": 3.4200000000000003e-06,
      "loss": 7.1984,
      "step": 61
    },
    {
      "epoch": 0.21,
      "learning_rate": 3.48e-06,
      "loss": 7.1072,
      "step": 62
    },
    {
      "epoch": 0.21,
      "learning_rate": 3.54e-06,
      "loss": 7.2887,
      "step": 63
    },
    {
      "epoch": 0.22,
      "learning_rate": 3.6e-06,
      "loss": 7.2358,
      "step": 64
    },
    {
      "epoch": 0.22,
      "learning_rate": 3.66e-06,
      "loss": 7.2417,
      "step": 65
    },
    {
      "epoch": 0.22,
      "learning_rate": 3.72e-06,
      "loss": 7.2254,
      "step": 66
    },
    {
      "epoch": 0.23,
      "learning_rate": 3.7800000000000002e-06,
      "loss": 7.2706,
      "step": 67
    },
    {
      "epoch": 0.23,
      "learning_rate": 3.8400000000000005e-06,
      "loss": 7.1927,
      "step": 68
    },
    {
      "epoch": 0.23,
      "learning_rate": 3.9e-06,
      "loss": 7.2085,
      "step": 69
    },
    {
      "epoch": 0.24,
      "learning_rate": 3.96e-06,
      "loss": 7.2298,
      "step": 70
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.0200000000000005e-06,
      "loss": 7.156,
      "step": 71
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.080000000000001e-06,
      "loss": 7.2131,
      "step": 72
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.14e-06,
      "loss": 7.2013,
      "step": 73
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.2000000000000004e-06,
      "loss": 7.1922,
      "step": 74
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.26e-06,
      "loss": 7.2448,
      "step": 75
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.32e-06,
      "loss": 7.1911,
      "step": 76
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.3799999999999996e-06,
      "loss": 7.1303,
      "step": 77
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.44e-06,
      "loss": 7.2067,
      "step": 78
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.5e-06,
      "loss": 7.1918,
      "step": 79
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.56e-06,
      "loss": 7.2269,
      "step": 80
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.62e-06,
      "loss": 7.1346,
      "step": 81
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.68e-06,
      "loss": 7.2346,
      "step": 82
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.74e-06,
      "loss": 7.1372,
      "step": 83
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.800000000000001e-06,
      "loss": 7.194,
      "step": 84
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.86e-06,
      "loss": 7.2357,
      "step": 85
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.92e-06,
      "loss": 7.1294,
      "step": 86
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.980000000000001e-06,
      "loss": 7.1368,
      "step": 87
    },
    {
      "epoch": 0.3,
      "learning_rate": 5.04e-06,
      "loss": 7.1945,
      "step": 88
    },
    {
      "epoch": 0.3,
      "learning_rate": 5.1e-06,
      "loss": 7.117,
      "step": 89
    },
    {
      "epoch": 0.3,
      "learning_rate": 5.16e-06,
      "loss": 7.2208,
      "step": 90
    },
    {
      "epoch": 0.31,
      "learning_rate": 5.22e-06,
      "loss": 7.1461,
      "step": 91
    },
    {
      "epoch": 0.31,
      "learning_rate": 5.279999999999999e-06,
      "loss": 7.3041,
      "step": 92
    },
    {
      "epoch": 0.31,
      "learning_rate": 5.34e-06,
      "loss": 7.1326,
      "step": 93
    },
    {
      "epoch": 0.32,
      "learning_rate": 5.4e-06,
      "loss": 7.0748,
      "step": 94
    },
    {
      "epoch": 0.32,
      "learning_rate": 5.46e-06,
      "loss": 6.9675,
      "step": 95
    },
    {
      "epoch": 0.32,
      "learning_rate": 5.52e-06,
      "loss": 7.1709,
      "step": 96
    },
    {
      "epoch": 0.33,
      "learning_rate": 5.58e-06,
      "loss": 7.0904,
      "step": 97
    },
    {
      "epoch": 0.33,
      "learning_rate": 5.64e-06,
      "loss": 7.0292,
      "step": 98
    },
    {
      "epoch": 0.33,
      "learning_rate": 5.7000000000000005e-06,
      "loss": 6.8978,
      "step": 99
    },
    {
      "epoch": 0.34,
      "learning_rate": 5.76e-06,
      "loss": 6.7386,
      "step": 100
    },
    {
      "epoch": 0.34,
      "learning_rate": 5.82e-06,
      "loss": 7.2255,
      "step": 101
    },
    {
      "epoch": 0.34,
      "learning_rate": 5.8800000000000005e-06,
      "loss": 7.2272,
      "step": 102
    },
    {
      "epoch": 0.35,
      "learning_rate": 5.940000000000001e-06,
      "loss": 7.1543,
      "step": 103
    },
    {
      "epoch": 0.35,
      "learning_rate": 6e-06,
      "loss": 7.1046,
      "step": 104
    },
    {
      "epoch": 0.35,
      "learning_rate": 6.0600000000000004e-06,
      "loss": 7.1356,
      "step": 105
    },
    {
      "epoch": 0.36,
      "learning_rate": 6.12e-06,
      "loss": 7.1201,
      "step": 106
    },
    {
      "epoch": 0.36,
      "learning_rate": 6.18e-06,
      "loss": 7.1179,
      "step": 107
    },
    {
      "epoch": 0.36,
      "learning_rate": 6.2399999999999995e-06,
      "loss": 7.1474,
      "step": 108
    },
    {
      "epoch": 0.37,
      "learning_rate": 6.3e-06,
      "loss": 7.1413,
      "step": 109
    },
    {
      "epoch": 0.37,
      "learning_rate": 6.36e-06,
      "loss": 7.148,
      "step": 110
    },
    {
      "epoch": 0.37,
      "learning_rate": 6.42e-06,
      "loss": 7.1286,
      "step": 111
    },
    {
      "epoch": 0.38,
      "learning_rate": 6.48e-06,
      "loss": 7.1098,
      "step": 112
    },
    {
      "epoch": 0.38,
      "learning_rate": 6.54e-06,
      "loss": 7.0648,
      "step": 113
    },
    {
      "epoch": 0.38,
      "learning_rate": 6.6e-06,
      "loss": 7.0697,
      "step": 114
    },
    {
      "epoch": 0.39,
      "learning_rate": 6.660000000000001e-06,
      "loss": 7.0634,
      "step": 115
    },
    {
      "epoch": 0.39,
      "learning_rate": 6.72e-06,
      "loss": 7.1039,
      "step": 116
    },
    {
      "epoch": 0.39,
      "learning_rate": 6.78e-06,
      "loss": 7.0493,
      "step": 117
    },
    {
      "epoch": 0.4,
      "learning_rate": 6.840000000000001e-06,
      "loss": 7.0045,
      "step": 118
    },
    {
      "epoch": 0.4,
      "learning_rate": 6.900000000000001e-06,
      "loss": 7.06,
      "step": 119
    },
    {
      "epoch": 0.4,
      "learning_rate": 6.96e-06,
      "loss": 7.1162,
      "step": 120
    },
    {
      "epoch": 0.41,
      "learning_rate": 7.0200000000000006e-06,
      "loss": 7.1069,
      "step": 121
    },
    {
      "epoch": 0.41,
      "learning_rate": 7.08e-06,
      "loss": 7.1361,
      "step": 122
    },
    {
      "epoch": 0.41,
      "learning_rate": 7.14e-06,
      "loss": 7.0522,
      "step": 123
    },
    {
      "epoch": 0.42,
      "learning_rate": 7.2e-06,
      "loss": 7.1425,
      "step": 124
    },
    {
      "epoch": 0.42,
      "learning_rate": 7.26e-06,
      "loss": 7.122,
      "step": 125
    },
    {
      "epoch": 0.42,
      "learning_rate": 7.32e-06,
      "loss": 7.0333,
      "step": 126
    },
    {
      "epoch": 0.43,
      "learning_rate": 7.3800000000000005e-06,
      "loss": 7.1291,
      "step": 127
    },
    {
      "epoch": 0.43,
      "learning_rate": 7.44e-06,
      "loss": 7.031,
      "step": 128
    },
    {
      "epoch": 0.43,
      "learning_rate": 7.5e-06,
      "loss": 7.0702,
      "step": 129
    },
    {
      "epoch": 0.44,
      "learning_rate": 7.5600000000000005e-06,
      "loss": 7.1561,
      "step": 130
    },
    {
      "epoch": 0.44,
      "learning_rate": 7.62e-06,
      "loss": 7.153,
      "step": 131
    },
    {
      "epoch": 0.44,
      "learning_rate": 7.680000000000001e-06,
      "loss": 6.9336,
      "step": 132
    },
    {
      "epoch": 0.45,
      "learning_rate": 7.74e-06,
      "loss": 7.0843,
      "step": 133
    },
    {
      "epoch": 0.45,
      "learning_rate": 7.8e-06,
      "loss": 7.0564,
      "step": 134
    },
    {
      "epoch": 0.45,
      "learning_rate": 7.860000000000001e-06,
      "loss": 7.1525,
      "step": 135
    },
    {
      "epoch": 0.46,
      "learning_rate": 7.92e-06,
      "loss": 7.0617,
      "step": 136
    },
    {
      "epoch": 0.46,
      "learning_rate": 7.98e-06,
      "loss": 7.0104,
      "step": 137
    },
    {
      "epoch": 0.46,
      "learning_rate": 8.040000000000001e-06,
      "loss": 7.0098,
      "step": 138
    },
    {
      "epoch": 0.47,
      "learning_rate": 8.1e-06,
      "loss": 7.0288,
      "step": 139
    },
    {
      "epoch": 0.47,
      "learning_rate": 8.160000000000001e-06,
      "loss": 7.081,
      "step": 140
    },
    {
      "epoch": 0.47,
      "learning_rate": 8.220000000000001e-06,
      "loss": 7.2109,
      "step": 141
    },
    {
      "epoch": 0.48,
      "learning_rate": 8.28e-06,
      "loss": 7.0495,
      "step": 142
    },
    {
      "epoch": 0.48,
      "learning_rate": 8.340000000000001e-06,
      "loss": 7.0757,
      "step": 143
    },
    {
      "epoch": 0.48,
      "learning_rate": 8.400000000000001e-06,
      "loss": 7.0424,
      "step": 144
    },
    {
      "epoch": 0.49,
      "learning_rate": 8.459999999999999e-06,
      "loss": 6.9206,
      "step": 145
    },
    {
      "epoch": 0.49,
      "learning_rate": 8.52e-06,
      "loss": 6.8338,
      "step": 146
    },
    {
      "epoch": 0.49,
      "learning_rate": 8.58e-06,
      "loss": 6.8781,
      "step": 147
    },
    {
      "epoch": 0.5,
      "learning_rate": 8.64e-06,
      "loss": 6.9345,
      "step": 148
    },
    {
      "epoch": 0.5,
      "learning_rate": 8.7e-06,
      "loss": 6.7973,
      "step": 149
    },
    {
      "epoch": 0.5,
      "learning_rate": 8.759999999999999e-06,
      "loss": 6.6469,
      "step": 150
    },
    {
      "epoch": 0.51,
      "learning_rate": 8.82e-06,
      "loss": 7.2513,
      "step": 151
    },
    {
      "epoch": 0.51,
      "learning_rate": 8.88e-06,
      "loss": 7.1988,
      "step": 152
    },
    {
      "epoch": 0.51,
      "learning_rate": 8.939999999999999e-06,
      "loss": 7.209,
      "step": 153
    },
    {
      "epoch": 0.52,
      "learning_rate": 9e-06,
      "loss": 7.1933,
      "step": 154
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.06e-06,
      "loss": 7.1361,
      "step": 155
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.12e-06,
      "loss": 7.1474,
      "step": 156
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.18e-06,
      "loss": 7.0345,
      "step": 157
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.24e-06,
      "loss": 7.1172,
      "step": 158
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.3e-06,
      "loss": 7.0408,
      "step": 159
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.36e-06,
      "loss": 7.0592,
      "step": 160
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.42e-06,
      "loss": 7.0444,
      "step": 161
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.48e-06,
      "loss": 7.0639,
      "step": 162
    },
    {
      "epoch": 0.55,
      "learning_rate": 9.54e-06,
      "loss": 7.079,
      "step": 163
    },
    {
      "epoch": 0.55,
      "learning_rate": 9.600000000000001e-06,
      "loss": 7.0526,
      "step": 164
    },
    {
      "epoch": 0.55,
      "learning_rate": 9.66e-06,
      "loss": 7.0755,
      "step": 165
    },
    {
      "epoch": 0.56,
      "learning_rate": 9.72e-06,
      "loss": 6.9862,
      "step": 166
    },
    {
      "epoch": 0.56,
      "learning_rate": 9.780000000000001e-06,
      "loss": 7.0277,
      "step": 167
    },
    {
      "epoch": 0.56,
      "learning_rate": 9.84e-06,
      "loss": 7.0504,
      "step": 168
    },
    {
      "epoch": 0.57,
      "learning_rate": 9.9e-06,
      "loss": 7.0658,
      "step": 169
    },
    {
      "epoch": 0.57,
      "learning_rate": 9.960000000000001e-06,
      "loss": 7.0983,
      "step": 170
    },
    {
      "epoch": 0.58,
      "learning_rate": 1.002e-05,
      "loss": 7.0216,
      "step": 171
    },
    {
      "epoch": 0.58,
      "learning_rate": 1.008e-05,
      "loss": 7.0129,
      "step": 172
    },
    {
      "epoch": 0.58,
      "learning_rate": 1.0140000000000001e-05,
      "loss": 6.9565,
      "step": 173
    },
    {
      "epoch": 0.59,
      "learning_rate": 1.02e-05,
      "loss": 6.8776,
      "step": 174
    },
    {
      "epoch": 0.59,
      "learning_rate": 1.0260000000000002e-05,
      "loss": 7.1386,
      "step": 175
    },
    {
      "epoch": 0.59,
      "learning_rate": 1.032e-05,
      "loss": 7.0078,
      "step": 176
    },
    {
      "epoch": 0.6,
      "learning_rate": 1.0379999999999999e-05,
      "loss": 7.0659,
      "step": 177
    },
    {
      "epoch": 0.6,
      "learning_rate": 1.044e-05,
      "loss": 7.048,
      "step": 178
    },
    {
      "epoch": 0.6,
      "learning_rate": 1.05e-05,
      "loss": 7.0641,
      "step": 179
    },
    {
      "epoch": 0.61,
      "learning_rate": 1.0559999999999999e-05,
      "loss": 7.0902,
      "step": 180
    },
    {
      "epoch": 0.61,
      "learning_rate": 1.062e-05,
      "loss": 6.9962,
      "step": 181
    },
    {
      "epoch": 0.61,
      "learning_rate": 1.068e-05,
      "loss": 7.0124,
      "step": 182
    },
    {
      "epoch": 0.62,
      "learning_rate": 1.074e-05,
      "loss": 6.9452,
      "step": 183
    },
    {
      "epoch": 0.62,
      "learning_rate": 1.08e-05,
      "loss": 7.0435,
      "step": 184
    },
    {
      "epoch": 0.62,
      "learning_rate": 1.086e-05,
      "loss": 7.0237,
      "step": 185
    },
    {
      "epoch": 0.63,
      "learning_rate": 1.092e-05,
      "loss": 7.0518,
      "step": 186
    },
    {
      "epoch": 0.63,
      "learning_rate": 1.098e-05,
      "loss": 7.0806,
      "step": 187
    },
    {
      "epoch": 0.63,
      "learning_rate": 1.104e-05,
      "loss": 7.0405,
      "step": 188
    },
    {
      "epoch": 0.64,
      "learning_rate": 1.11e-05,
      "loss": 7.0664,
      "step": 189
    },
    {
      "epoch": 0.64,
      "learning_rate": 1.116e-05,
      "loss": 7.1205,
      "step": 190
    },
    {
      "epoch": 0.64,
      "learning_rate": 1.1220000000000001e-05,
      "loss": 7.0165,
      "step": 191
    },
    {
      "epoch": 0.65,
      "learning_rate": 1.128e-05,
      "loss": 7.0557,
      "step": 192
    },
    {
      "epoch": 0.65,
      "learning_rate": 1.134e-05,
      "loss": 6.9645,
      "step": 193
    },
    {
      "epoch": 0.65,
      "learning_rate": 1.1400000000000001e-05,
      "loss": 6.9475,
      "step": 194
    },
    {
      "epoch": 0.66,
      "learning_rate": 1.146e-05,
      "loss": 6.9141,
      "step": 195
    },
    {
      "epoch": 0.66,
      "learning_rate": 1.152e-05,
      "loss": 6.8144,
      "step": 196
    },
    {
      "epoch": 0.66,
      "learning_rate": 1.1580000000000001e-05,
      "loss": 6.8174,
      "step": 197
    },
    {
      "epoch": 0.67,
      "learning_rate": 1.164e-05,
      "loss": 6.9932,
      "step": 198
    },
    {
      "epoch": 0.67,
      "learning_rate": 1.1700000000000001e-05,
      "loss": 6.7158,
      "step": 199
    },
    {
      "epoch": 0.67,
      "learning_rate": 1.1760000000000001e-05,
      "loss": 6.5511,
      "step": 200
    },
    {
      "epoch": 0.68,
      "learning_rate": 1.182e-05,
      "loss": 7.3286,
      "step": 201
    },
    {
      "epoch": 0.68,
      "learning_rate": 1.1880000000000001e-05,
      "loss": 7.3258,
      "step": 202
    },
    {
      "epoch": 0.68,
      "learning_rate": 1.1940000000000001e-05,
      "loss": 7.2445,
      "step": 203
    },
    {
      "epoch": 0.69,
      "learning_rate": 1.2e-05,
      "loss": 7.1024,
      "step": 204
    },
    {
      "epoch": 0.69,
      "learning_rate": 1.2060000000000001e-05,
      "loss": 7.1517,
      "step": 205
    },
    {
      "epoch": 0.69,
      "learning_rate": 1.2120000000000001e-05,
      "loss": 7.0302,
      "step": 206
    },
    {
      "epoch": 0.7,
      "learning_rate": 1.2180000000000002e-05,
      "loss": 6.9979,
      "step": 207
    },
    {
      "epoch": 0.7,
      "learning_rate": 1.224e-05,
      "loss": 7.162,
      "step": 208
    },
    {
      "epoch": 0.7,
      "learning_rate": 1.2299999999999999e-05,
      "loss": 6.9514,
      "step": 209
    },
    {
      "epoch": 0.71,
      "learning_rate": 1.236e-05,
      "loss": 7.0485,
      "step": 210
    },
    {
      "epoch": 0.71,
      "learning_rate": 1.242e-05,
      "loss": 7.0077,
      "step": 211
    },
    {
      "epoch": 0.71,
      "learning_rate": 1.2479999999999999e-05,
      "loss": 6.964,
      "step": 212
    },
    {
      "epoch": 0.72,
      "learning_rate": 1.254e-05,
      "loss": 6.9713,
      "step": 213
    },
    {
      "epoch": 0.72,
      "learning_rate": 1.26e-05,
      "loss": 6.9846,
      "step": 214
    },
    {
      "epoch": 0.72,
      "learning_rate": 1.2659999999999999e-05,
      "loss": 7.0551,
      "step": 215
    },
    {
      "epoch": 0.73,
      "learning_rate": 1.272e-05,
      "loss": 7.1107,
      "step": 216
    },
    {
      "epoch": 0.73,
      "learning_rate": 1.278e-05,
      "loss": 7.0144,
      "step": 217
    },
    {
      "epoch": 0.73,
      "learning_rate": 1.284e-05,
      "loss": 6.922,
      "step": 218
    },
    {
      "epoch": 0.74,
      "learning_rate": 1.29e-05,
      "loss": 6.9366,
      "step": 219
    },
    {
      "epoch": 0.74,
      "learning_rate": 1.296e-05,
      "loss": 6.9862,
      "step": 220
    },
    {
      "epoch": 0.74,
      "learning_rate": 1.302e-05,
      "loss": 6.9755,
      "step": 221
    },
    {
      "epoch": 0.75,
      "learning_rate": 1.308e-05,
      "loss": 6.9939,
      "step": 222
    },
    {
      "epoch": 0.75,
      "learning_rate": 1.314e-05,
      "loss": 6.9933,
      "step": 223
    },
    {
      "epoch": 0.75,
      "learning_rate": 1.32e-05,
      "loss": 6.9913,
      "step": 224
    },
    {
      "epoch": 0.76,
      "learning_rate": 1.326e-05,
      "loss": 7.0128,
      "step": 225
    },
    {
      "epoch": 0.76,
      "learning_rate": 1.3320000000000001e-05,
      "loss": 7.0123,
      "step": 226
    },
    {
      "epoch": 0.76,
      "learning_rate": 1.338e-05,
      "loss": 7.024,
      "step": 227
    },
    {
      "epoch": 0.77,
      "learning_rate": 1.344e-05,
      "loss": 6.923,
      "step": 228
    },
    {
      "epoch": 0.77,
      "learning_rate": 1.3500000000000001e-05,
      "loss": 7.0585,
      "step": 229
    },
    {
      "epoch": 0.77,
      "learning_rate": 1.356e-05,
      "loss": 6.916,
      "step": 230
    },
    {
      "epoch": 0.78,
      "learning_rate": 1.362e-05,
      "loss": 7.0473,
      "step": 231
    },
    {
      "epoch": 0.78,
      "learning_rate": 1.3680000000000001e-05,
      "loss": 6.9184,
      "step": 232
    },
    {
      "epoch": 0.78,
      "learning_rate": 1.374e-05,
      "loss": 7.0242,
      "step": 233
    },
    {
      "epoch": 0.79,
      "learning_rate": 1.3800000000000002e-05,
      "loss": 6.9791,
      "step": 234
    },
    {
      "epoch": 0.79,
      "learning_rate": 1.3860000000000001e-05,
      "loss": 7.0786,
      "step": 235
    },
    {
      "epoch": 0.79,
      "learning_rate": 1.392e-05,
      "loss": 6.95,
      "step": 236
    },
    {
      "epoch": 0.8,
      "learning_rate": 1.3980000000000002e-05,
      "loss": 6.8634,
      "step": 237
    },
    {
      "epoch": 0.8,
      "learning_rate": 1.4040000000000001e-05,
      "loss": 7.0312,
      "step": 238
    },
    {
      "epoch": 0.8,
      "learning_rate": 1.4099999999999999e-05,
      "loss": 6.9264,
      "step": 239
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.416e-05,
      "loss": 6.896,
      "step": 240
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.422e-05,
      "loss": 6.912,
      "step": 241
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.428e-05,
      "loss": 6.9706,
      "step": 242
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.434e-05,
      "loss": 6.8722,
      "step": 243
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.44e-05,
      "loss": 6.7643,
      "step": 244
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.446e-05,
      "loss": 6.851,
      "step": 245
    },
    {
      "epoch": 0.83,
      "learning_rate": 1.452e-05,
      "loss": 6.8817,
      "step": 246
    },
    {
      "epoch": 0.83,
      "learning_rate": 1.458e-05,
      "loss": 6.7148,
      "step": 247
    },
    {
      "epoch": 0.83,
      "learning_rate": 1.464e-05,
      "loss": 6.5562,
      "step": 248
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.47e-05,
      "loss": 6.614,
      "step": 249
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.4760000000000001e-05,
      "loss": 6.3515,
      "step": 250
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.482e-05,
      "loss": 7.3691,
      "step": 251
    },
    {
      "epoch": 0.85,
      "learning_rate": 1.488e-05,
      "loss": 7.3315,
      "step": 252
    },
    {
      "epoch": 0.85,
      "learning_rate": 1.4940000000000001e-05,
      "loss": 7.3868,
      "step": 253
    },
    {
      "epoch": 0.85,
      "learning_rate": 1.5e-05,
      "loss": 7.2316,
      "step": 254
    },
    {
      "epoch": 0.86,
      "learning_rate": 1.506e-05,
      "loss": 7.0674,
      "step": 255
    },
    {
      "epoch": 0.86,
      "learning_rate": 1.5120000000000001e-05,
      "loss": 6.9861,
      "step": 256
    },
    {
      "epoch": 0.86,
      "learning_rate": 1.518e-05,
      "loss": 7.0043,
      "step": 257
    },
    {
      "epoch": 0.87,
      "learning_rate": 1.524e-05,
      "loss": 6.9465,
      "step": 258
    },
    {
      "epoch": 0.87,
      "learning_rate": 1.53e-05,
      "loss": 6.9224,
      "step": 259
    },
    {
      "epoch": 0.87,
      "learning_rate": 1.5360000000000002e-05,
      "loss": 6.8833,
      "step": 260
    },
    {
      "epoch": 0.88,
      "learning_rate": 1.542e-05,
      "loss": 6.9311,
      "step": 261
    },
    {
      "epoch": 0.88,
      "learning_rate": 1.548e-05,
      "loss": 6.9703,
      "step": 262
    },
    {
      "epoch": 0.88,
      "learning_rate": 1.554e-05,
      "loss": 6.8447,
      "step": 263
    },
    {
      "epoch": 0.89,
      "learning_rate": 1.56e-05,
      "loss": 6.9624,
      "step": 264
    },
    {
      "epoch": 0.89,
      "learning_rate": 1.5660000000000003e-05,
      "loss": 6.8964,
      "step": 265
    },
    {
      "epoch": 0.89,
      "learning_rate": 1.5720000000000002e-05,
      "loss": 6.9926,
      "step": 266
    },
    {
      "epoch": 0.9,
      "learning_rate": 1.578e-05,
      "loss": 6.911,
      "step": 267
    },
    {
      "epoch": 0.9,
      "learning_rate": 1.584e-05,
      "loss": 6.9273,
      "step": 268
    },
    {
      "epoch": 0.9,
      "learning_rate": 1.59e-05,
      "loss": 6.9688,
      "step": 269
    },
    {
      "epoch": 0.91,
      "learning_rate": 1.596e-05,
      "loss": 6.922,
      "step": 270
    },
    {
      "epoch": 0.91,
      "learning_rate": 1.6020000000000002e-05,
      "loss": 6.8924,
      "step": 271
    },
    {
      "epoch": 0.91,
      "learning_rate": 1.6080000000000002e-05,
      "loss": 6.8255,
      "step": 272
    },
    {
      "epoch": 0.92,
      "learning_rate": 1.614e-05,
      "loss": 6.9048,
      "step": 273
    },
    {
      "epoch": 0.92,
      "learning_rate": 1.62e-05,
      "loss": 6.8165,
      "step": 274
    },
    {
      "epoch": 0.92,
      "learning_rate": 1.626e-05,
      "loss": 6.9853,
      "step": 275
    },
    {
      "epoch": 0.93,
      "learning_rate": 1.6320000000000003e-05,
      "loss": 6.8935,
      "step": 276
    },
    {
      "epoch": 0.93,
      "learning_rate": 1.6380000000000002e-05,
      "loss": 6.9216,
      "step": 277
    },
    {
      "epoch": 0.93,
      "learning_rate": 1.6440000000000002e-05,
      "loss": 6.8888,
      "step": 278
    },
    {
      "epoch": 0.94,
      "learning_rate": 1.65e-05,
      "loss": 6.8371,
      "step": 279
    },
    {
      "epoch": 0.94,
      "learning_rate": 1.656e-05,
      "loss": 6.8125,
      "step": 280
    },
    {
      "epoch": 0.94,
      "learning_rate": 1.6620000000000004e-05,
      "loss": 6.8752,
      "step": 281
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.6680000000000003e-05,
      "loss": 6.8624,
      "step": 282
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.6740000000000002e-05,
      "loss": 6.9626,
      "step": 283
    },
    {
      "epoch": 0.96,
      "learning_rate": 1.6800000000000002e-05,
      "loss": 6.8377,
      "step": 284
    },
    {
      "epoch": 0.96,
      "learning_rate": 1.686e-05,
      "loss": 6.8724,
      "step": 285
    },
    {
      "epoch": 0.96,
      "learning_rate": 1.6919999999999997e-05,
      "loss": 6.9337,
      "step": 286
    },
    {
      "epoch": 0.97,
      "learning_rate": 1.698e-05,
      "loss": 6.8343,
      "step": 287
    },
    {
      "epoch": 0.97,
      "learning_rate": 1.704e-05,
      "loss": 6.7935,
      "step": 288
    },
    {
      "epoch": 0.97,
      "learning_rate": 1.71e-05,
      "loss": 6.8041,
      "step": 289
    },
    {
      "epoch": 0.98,
      "learning_rate": 1.716e-05,
      "loss": 6.8632,
      "step": 290
    },
    {
      "epoch": 0.98,
      "learning_rate": 1.7219999999999998e-05,
      "loss": 6.7646,
      "step": 291
    },
    {
      "epoch": 0.98,
      "learning_rate": 1.728e-05,
      "loss": 6.8201,
      "step": 292
    },
    {
      "epoch": 0.99,
      "learning_rate": 1.734e-05,
      "loss": 6.7473,
      "step": 293
    },
    {
      "epoch": 0.99,
      "learning_rate": 1.74e-05,
      "loss": 6.6929,
      "step": 294
    },
    {
      "epoch": 0.99,
      "learning_rate": 1.746e-05,
      "loss": 6.6229,
      "step": 295
    },
    {
      "epoch": 1.0,
      "learning_rate": 1.7519999999999998e-05,
      "loss": 6.7064,
      "step": 296
    },
    {
      "epoch": 1.0,
      "learning_rate": 1.758e-05,
      "loss": 6.3378,
      "step": 297
    },
    {
      "epoch": 1.0,
      "step": 297,
      "total_flos": 0.0,
      "train_loss": 7.210098851008046,
      "train_runtime": 4644.899,
      "train_samples_per_second": 6.144,
      "train_steps_per_second": 0.064
    }
  ],
  "max_steps": 297,
  "num_train_epochs": 1,
  "total_flos": 0.0,
  "trial_name": null,
  "trial_params": null
}