{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "global_step": 987,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "learning_rate": 0.0,
      "loss": 1.1206,
      "step": 1
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.075900941810124e-06,
      "loss": 0.9243,
      "step": 2
    },
    {
      "epoch": 0.01,
      "learning_rate": 6.46015014942309e-06,
      "loss": 0.9025,
      "step": 3
    },
    {
      "epoch": 0.01,
      "learning_rate": 8.151801883620247e-06,
      "loss": 0.7408,
      "step": 4
    },
    {
      "epoch": 0.02,
      "learning_rate": 9.463948908766788e-06,
      "loss": 0.6775,
      "step": 5
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.0536051091233212e-05,
      "loss": 0.5847,
      "step": 6
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.1442500570809876e-05,
      "loss": 0.5582,
      "step": 7
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.222770282543037e-05,
      "loss": 0.5626,
      "step": 8
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.292030029884618e-05,
      "loss": 0.5166,
      "step": 9
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.3539849850576912e-05,
      "loss": 0.5106,
      "step": 10
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.4100300592531481e-05,
      "loss": 0.4948,
      "step": 11
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.4611952033043337e-05,
      "loss": 0.4977,
      "step": 12
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.5082625732282867e-05,
      "loss": 0.4904,
      "step": 13
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.551840151262e-05,
      "loss": 0.4836,
      "step": 14
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.5924099058189875e-05,
      "loss": 0.4746,
      "step": 15
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.6303603767240495e-05,
      "loss": 0.4723,
      "step": 16
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.6660093644266146e-05,
      "loss": 0.4798,
      "step": 17
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.6996201240656302e-05,
      "loss": 0.4886,
      "step": 18
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.7314131752785847e-05,
      "loss": 0.47,
      "step": 19
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.7615750792387035e-05,
      "loss": 0.4566,
      "step": 20
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.7902650720232966e-05,
      "loss": 0.4755,
      "step": 21
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.8176201534341607e-05,
      "loss": 0.4553,
      "step": 22
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.8437590437029225e-05,
      "loss": 0.4454,
      "step": 23
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.868785297485346e-05,
      "loss": 0.4348,
      "step": 24
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.8927897817533575e-05,
      "loss": 0.445,
      "step": 25
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.915852667409299e-05,
      "loss": 0.434,
      "step": 26
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.9380450448269272e-05,
      "loss": 0.4453,
      "step": 27
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.9594302454430122e-05,
      "loss": 0.4324,
      "step": 28
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.9800649313336155e-05,
      "loss": 0.4442,
      "step": 29
    },
    {
      "epoch": 0.09,
      "learning_rate": 2e-05,
      "loss": 0.44,
      "step": 30
    },
    {
      "epoch": 0.09,
      "learning_rate": 2e-05,
      "loss": 0.4219,
      "step": 31
    },
    {
      "epoch": 0.1,
      "learning_rate": 2e-05,
      "loss": 0.4335,
      "step": 32
    },
    {
      "epoch": 0.1,
      "learning_rate": 2e-05,
      "loss": 0.4266,
      "step": 33
    },
    {
      "epoch": 0.1,
      "learning_rate": 2e-05,
      "loss": 0.4295,
      "step": 34
    },
    {
      "epoch": 0.11,
      "learning_rate": 2e-05,
      "loss": 0.4345,
      "step": 35
    },
    {
      "epoch": 0.11,
      "learning_rate": 2e-05,
      "loss": 0.4251,
      "step": 36
    },
    {
      "epoch": 0.11,
      "learning_rate": 2e-05,
      "loss": 0.4364,
      "step": 37
    },
    {
      "epoch": 0.12,
      "learning_rate": 2e-05,
      "loss": 0.429,
      "step": 38
    },
    {
      "epoch": 0.12,
      "learning_rate": 2e-05,
      "loss": 0.4222,
      "step": 39
    },
    {
      "epoch": 0.12,
      "learning_rate": 2e-05,
      "loss": 0.4164,
      "step": 40
    },
    {
      "epoch": 0.12,
      "learning_rate": 2e-05,
      "loss": 0.4141,
      "step": 41
    },
    {
      "epoch": 0.13,
      "learning_rate": 2e-05,
      "loss": 0.4324,
      "step": 42
    },
    {
      "epoch": 0.13,
      "learning_rate": 2e-05,
      "loss": 0.4061,
      "step": 43
    },
    {
      "epoch": 0.13,
      "learning_rate": 2e-05,
      "loss": 0.4233,
      "step": 44
    },
    {
      "epoch": 0.14,
      "learning_rate": 2e-05,
      "loss": 0.4014,
      "step": 45
    },
    {
      "epoch": 0.14,
      "learning_rate": 2e-05,
      "loss": 0.4033,
      "step": 46
    },
    {
      "epoch": 0.14,
      "learning_rate": 2e-05,
      "loss": 0.4043,
      "step": 47
    },
    {
      "epoch": 0.15,
      "learning_rate": 2e-05,
      "loss": 0.4131,
      "step": 48
    },
    {
      "epoch": 0.15,
      "learning_rate": 2e-05,
      "loss": 0.393,
      "step": 49
    },
    {
      "epoch": 0.15,
      "learning_rate": 2e-05,
      "loss": 0.3902,
      "step": 50
    },
    {
      "epoch": 0.16,
      "learning_rate": 2e-05,
      "loss": 0.401,
      "step": 51
    },
    {
      "epoch": 0.16,
      "learning_rate": 2e-05,
      "loss": 0.383,
      "step": 52
    },
    {
      "epoch": 0.16,
      "learning_rate": 2e-05,
      "loss": 0.3885,
      "step": 53
    },
    {
      "epoch": 0.16,
      "learning_rate": 2e-05,
      "loss": 0.3947,
      "step": 54
    },
    {
      "epoch": 0.17,
      "learning_rate": 2e-05,
      "loss": 0.3914,
      "step": 55
    },
    {
      "epoch": 0.17,
      "learning_rate": 2e-05,
      "loss": 0.379,
      "step": 56
    },
    {
      "epoch": 0.17,
      "learning_rate": 2e-05,
      "loss": 0.4084,
      "step": 57
    },
    {
      "epoch": 0.18,
      "learning_rate": 2e-05,
      "loss": 0.3933,
      "step": 58
    },
    {
      "epoch": 0.18,
      "learning_rate": 2e-05,
      "loss": 0.396,
      "step": 59
    },
    {
      "epoch": 0.18,
      "learning_rate": 2e-05,
      "loss": 0.3976,
      "step": 60
    },
    {
      "epoch": 0.19,
      "learning_rate": 2e-05,
      "loss": 0.3934,
      "step": 61
    },
    {
      "epoch": 0.19,
      "learning_rate": 2e-05,
      "loss": 0.3938,
      "step": 62
    },
    {
      "epoch": 0.19,
      "learning_rate": 2e-05,
      "loss": 0.3882,
      "step": 63
    },
    {
      "epoch": 0.19,
      "learning_rate": 2e-05,
      "loss": 0.3865,
      "step": 64
    },
    {
      "epoch": 0.2,
      "learning_rate": 2e-05,
      "loss": 0.3929,
      "step": 65
    },
    {
      "epoch": 0.2,
      "learning_rate": 2e-05,
      "loss": 0.3895,
      "step": 66
    },
    {
      "epoch": 0.2,
      "learning_rate": 2e-05,
      "loss": 0.3754,
      "step": 67
    },
    {
      "epoch": 0.21,
      "learning_rate": 2e-05,
      "loss": 0.4001,
      "step": 68
    },
    {
      "epoch": 0.21,
      "learning_rate": 2e-05,
      "loss": 0.3873,
      "step": 69
    },
    {
      "epoch": 0.21,
      "learning_rate": 2e-05,
      "loss": 0.3829,
      "step": 70
    },
    {
      "epoch": 0.22,
      "learning_rate": 2e-05,
      "loss": 0.3988,
      "step": 71
    },
    {
      "epoch": 0.22,
      "learning_rate": 2e-05,
      "loss": 0.3787,
      "step": 72
    },
    {
      "epoch": 0.22,
      "learning_rate": 2e-05,
      "loss": 0.3835,
      "step": 73
    },
    {
      "epoch": 0.22,
      "learning_rate": 2e-05,
      "loss": 0.376,
      "step": 74
    },
    {
      "epoch": 0.23,
      "learning_rate": 2e-05,
      "loss": 0.3765,
      "step": 75
    },
    {
      "epoch": 0.23,
      "learning_rate": 2e-05,
      "loss": 0.3813,
      "step": 76
    },
    {
      "epoch": 0.23,
      "learning_rate": 2e-05,
      "loss": 0.3792,
      "step": 77
    },
    {
      "epoch": 0.24,
      "learning_rate": 2e-05,
      "loss": 0.3822,
      "step": 78
    },
    {
      "epoch": 0.24,
      "learning_rate": 2e-05,
      "loss": 0.3799,
      "step": 79
    },
    {
      "epoch": 0.24,
      "learning_rate": 2e-05,
      "loss": 0.3666,
      "step": 80
    },
    {
      "epoch": 0.25,
      "learning_rate": 2e-05,
      "loss": 0.3938,
      "step": 81
    },
    {
      "epoch": 0.25,
      "learning_rate": 2e-05,
      "loss": 0.3868,
      "step": 82
    },
    {
      "epoch": 0.25,
      "learning_rate": 2e-05,
      "loss": 0.377,
      "step": 83
    },
    {
      "epoch": 0.26,
      "learning_rate": 2e-05,
      "loss": 0.3681,
      "step": 84
    },
    {
      "epoch": 0.26,
      "learning_rate": 2e-05,
      "loss": 0.3838,
      "step": 85
    },
    {
      "epoch": 0.26,
      "learning_rate": 2e-05,
      "loss": 0.3867,
      "step": 86
    },
    {
      "epoch": 0.26,
      "learning_rate": 2e-05,
      "loss": 0.3845,
      "step": 87
    },
    {
      "epoch": 0.27,
      "learning_rate": 2e-05,
      "loss": 0.3761,
      "step": 88
    },
    {
      "epoch": 0.27,
      "learning_rate": 2e-05,
      "loss": 0.3695,
      "step": 89
    },
    {
      "epoch": 0.27,
      "learning_rate": 2e-05,
      "loss": 0.39,
      "step": 90
    },
    {
      "epoch": 0.28,
      "learning_rate": 2e-05,
      "loss": 0.366,
      "step": 91
    },
    {
      "epoch": 0.28,
      "learning_rate": 2e-05,
      "loss": 0.3876,
      "step": 92
    },
    {
      "epoch": 0.28,
      "learning_rate": 2e-05,
      "loss": 0.3727,
      "step": 93
    },
    {
      "epoch": 0.29,
      "learning_rate": 2e-05,
      "loss": 0.382,
      "step": 94
    },
    {
      "epoch": 0.29,
      "learning_rate": 2e-05,
      "loss": 0.3721,
      "step": 95
    },
    {
      "epoch": 0.29,
      "learning_rate": 2e-05,
      "loss": 0.3736,
      "step": 96
    },
    {
      "epoch": 0.29,
      "learning_rate": 2e-05,
      "loss": 0.3819,
      "step": 97
    },
    {
      "epoch": 0.3,
      "learning_rate": 2e-05,
      "loss": 0.3852,
      "step": 98
    },
    {
      "epoch": 0.3,
      "learning_rate": 2e-05,
      "loss": 0.3645,
      "step": 99
    },
    {
      "epoch": 0.3,
      "learning_rate": 2e-05,
      "loss": 0.3643,
      "step": 100
    },
    {
      "epoch": 0.31,
      "learning_rate": 2e-05,
      "loss": 0.3774,
      "step": 101
    },
    {
      "epoch": 0.31,
      "learning_rate": 2e-05,
      "loss": 0.3822,
      "step": 102
    },
    {
      "epoch": 0.31,
      "learning_rate": 2e-05,
      "loss": 0.361,
      "step": 103
    },
    {
      "epoch": 0.32,
      "learning_rate": 2e-05,
      "loss": 0.3832,
      "step": 104
    },
    {
      "epoch": 0.32,
      "learning_rate": 2e-05,
      "loss": 0.3655,
      "step": 105
    },
    {
      "epoch": 0.32,
      "learning_rate": 2e-05,
      "loss": 0.3758,
      "step": 106
    },
    {
      "epoch": 0.33,
      "learning_rate": 2e-05,
      "loss": 0.3556,
      "step": 107
    },
    {
      "epoch": 0.33,
      "learning_rate": 2e-05,
      "loss": 0.3813,
      "step": 108
    },
    {
      "epoch": 0.33,
      "learning_rate": 2e-05,
      "loss": 0.3681,
      "step": 109
    },
    {
      "epoch": 0.33,
      "learning_rate": 2e-05,
      "loss": 0.3706,
      "step": 110
    },
    {
      "epoch": 0.34,
      "learning_rate": 2e-05,
      "loss": 0.3889,
      "step": 111
    },
    {
      "epoch": 0.34,
      "learning_rate": 2e-05,
      "loss": 0.3759,
      "step": 112
    },
    {
      "epoch": 0.34,
      "learning_rate": 2e-05,
      "loss": 0.3604,
      "step": 113
    },
    {
      "epoch": 0.35,
      "learning_rate": 2e-05,
      "loss": 0.3746,
      "step": 114
    },
    {
      "epoch": 0.35,
      "learning_rate": 2e-05,
      "loss": 0.3678,
      "step": 115
    },
    {
      "epoch": 0.35,
      "learning_rate": 2e-05,
      "loss": 0.3784,
      "step": 116
    },
    {
      "epoch": 0.36,
      "learning_rate": 2e-05,
      "loss": 0.3687,
      "step": 117
    },
    {
      "epoch": 0.36,
      "learning_rate": 2e-05,
      "loss": 0.3743,
      "step": 118
    },
    {
      "epoch": 0.36,
      "learning_rate": 2e-05,
      "loss": 0.378,
      "step": 119
    },
    {
      "epoch": 0.36,
      "learning_rate": 2e-05,
      "loss": 0.3675,
      "step": 120
    },
    {
      "epoch": 0.37,
      "learning_rate": 2e-05,
      "loss": 0.3664,
      "step": 121
    },
    {
      "epoch": 0.37,
      "learning_rate": 2e-05,
      "loss": 0.3861,
      "step": 122
    },
    {
      "epoch": 0.37,
      "learning_rate": 2e-05,
      "loss": 0.3618,
      "step": 123
    },
    {
      "epoch": 0.38,
      "learning_rate": 2e-05,
      "loss": 0.3906,
      "step": 124
    },
    {
      "epoch": 0.38,
      "learning_rate": 2e-05,
      "loss": 0.3613,
      "step": 125
    },
    {
      "epoch": 0.38,
      "learning_rate": 2e-05,
      "loss": 0.3756,
      "step": 126
    },
    {
      "epoch": 0.39,
      "learning_rate": 2e-05,
      "loss": 0.3652,
      "step": 127
    },
    {
      "epoch": 0.39,
      "learning_rate": 2e-05,
      "loss": 0.3618,
      "step": 128
    },
    {
      "epoch": 0.39,
      "learning_rate": 2e-05,
      "loss": 0.3634,
      "step": 129
    },
    {
      "epoch": 0.4,
      "learning_rate": 2e-05,
      "loss": 0.3684,
      "step": 130
    },
    {
      "epoch": 0.4,
      "learning_rate": 2e-05,
      "loss": 0.3844,
      "step": 131
    },
    {
      "epoch": 0.4,
      "learning_rate": 2e-05,
      "loss": 0.3563,
      "step": 132
    },
    {
      "epoch": 0.4,
      "learning_rate": 2e-05,
      "loss": 0.3697,
      "step": 133
    },
    {
      "epoch": 0.41,
      "learning_rate": 2e-05,
      "loss": 0.3636,
      "step": 134
    },
    {
      "epoch": 0.41,
      "learning_rate": 2e-05,
      "loss": 0.3629,
      "step": 135
    },
    {
      "epoch": 0.41,
      "learning_rate": 2e-05,
      "loss": 0.3744,
      "step": 136
    },
    {
      "epoch": 0.42,
      "learning_rate": 2e-05,
      "loss": 0.3483,
      "step": 137
    },
    {
      "epoch": 0.42,
      "learning_rate": 2e-05,
      "loss": 0.3618,
      "step": 138
    },
    {
      "epoch": 0.42,
      "learning_rate": 2e-05,
      "loss": 0.3612,
      "step": 139
    },
    {
      "epoch": 0.43,
      "learning_rate": 2e-05,
      "loss": 0.3623,
      "step": 140
    },
    {
      "epoch": 0.43,
      "learning_rate": 2e-05,
      "loss": 0.3633,
      "step": 141
    },
    {
      "epoch": 0.43,
      "learning_rate": 2e-05,
      "loss": 0.3688,
      "step": 142
    },
    {
      "epoch": 0.43,
      "learning_rate": 2e-05,
      "loss": 0.3618,
      "step": 143
    },
    {
      "epoch": 0.44,
      "learning_rate": 2e-05,
      "loss": 0.3654,
      "step": 144
    },
    {
      "epoch": 0.44,
      "learning_rate": 2e-05,
      "loss": 0.3787,
      "step": 145
    },
    {
      "epoch": 0.44,
      "learning_rate": 2e-05,
      "loss": 0.383,
      "step": 146
    },
    {
      "epoch": 0.45,
      "learning_rate": 2e-05,
      "loss": 0.3639,
      "step": 147
    },
    {
      "epoch": 0.45,
      "learning_rate": 2e-05,
      "loss": 0.3725,
      "step": 148
    },
    {
      "epoch": 0.45,
      "learning_rate": 2e-05,
      "loss": 0.3553,
      "step": 149
    },
    {
      "epoch": 0.46,
      "learning_rate": 2e-05,
      "loss": 0.3687,
      "step": 150
    },
    {
      "epoch": 0.46,
      "learning_rate": 2e-05,
      "loss": 0.371,
      "step": 151
    },
    {
      "epoch": 0.46,
      "learning_rate": 2e-05,
      "loss": 0.3677,
      "step": 152
    },
    {
      "epoch": 0.47,
      "learning_rate": 2e-05,
      "loss": 0.3539,
      "step": 153
    },
    {
      "epoch": 0.47,
      "learning_rate": 2e-05,
      "loss": 0.3729,
      "step": 154
    },
    {
      "epoch": 0.47,
      "learning_rate": 2e-05,
      "loss": 0.3637,
      "step": 155
    },
    {
      "epoch": 0.47,
      "learning_rate": 2e-05,
      "loss": 0.3604,
      "step": 156
    },
    {
      "epoch": 0.48,
      "learning_rate": 2e-05,
      "loss": 0.346,
      "step": 157
    },
    {
      "epoch": 0.48,
      "learning_rate": 2e-05,
      "loss": 0.3679,
      "step": 158
    },
    {
      "epoch": 0.48,
      "learning_rate": 2e-05,
      "loss": 0.3659,
      "step": 159
    },
    {
      "epoch": 0.49,
      "learning_rate": 2e-05,
      "loss": 0.3592,
      "step": 160
    },
    {
      "epoch": 0.49,
      "learning_rate": 2e-05,
      "loss": 0.3514,
      "step": 161
    },
    {
      "epoch": 0.49,
      "learning_rate": 2e-05,
      "loss": 0.3774,
      "step": 162
    },
    {
      "epoch": 0.5,
      "learning_rate": 2e-05,
      "loss": 0.3696,
      "step": 163
    },
    {
      "epoch": 0.5,
      "learning_rate": 2e-05,
      "loss": 0.3558,
      "step": 164
    },
    {
      "epoch": 0.5,
      "learning_rate": 2e-05,
      "loss": 0.3757,
      "step": 165
    },
    {
      "epoch": 0.5,
      "learning_rate": 2e-05,
      "loss": 0.3643,
      "step": 166
    },
    {
      "epoch": 0.51,
      "learning_rate": 2e-05,
      "loss": 0.3609,
      "step": 167
    },
    {
      "epoch": 0.51,
      "learning_rate": 2e-05,
      "loss": 0.3545,
      "step": 168
    },
    {
      "epoch": 0.51,
      "learning_rate": 2e-05,
      "loss": 0.3634,
      "step": 169
    },
    {
      "epoch": 0.52,
      "learning_rate": 2e-05,
      "loss": 0.3615,
      "step": 170
    },
    {
      "epoch": 0.52,
      "learning_rate": 2e-05,
      "loss": 0.3577,
      "step": 171
    },
    {
      "epoch": 0.52,
      "learning_rate": 2e-05,
      "loss": 0.3422,
      "step": 172
    },
    {
      "epoch": 0.53,
      "learning_rate": 2e-05,
      "loss": 0.3664,
      "step": 173
    },
    {
      "epoch": 0.53,
      "learning_rate": 2e-05,
      "loss": 0.3613,
      "step": 174
    },
    {
      "epoch": 0.53,
      "learning_rate": 2e-05,
      "loss": 0.3654,
      "step": 175
    },
    {
      "epoch": 0.53,
      "learning_rate": 2e-05,
      "loss": 0.3631,
      "step": 176
    },
    {
      "epoch": 0.54,
      "learning_rate": 2e-05,
      "loss": 0.3651,
      "step": 177
    },
    {
      "epoch": 0.54,
      "learning_rate": 2e-05,
      "loss": 0.362,
      "step": 178
    },
    {
      "epoch": 0.54,
      "learning_rate": 2e-05,
      "loss": 0.3792,
      "step": 179
    },
    {
      "epoch": 0.55,
      "learning_rate": 2e-05,
      "loss": 0.3643,
      "step": 180
    },
    {
      "epoch": 0.55,
      "learning_rate": 2e-05,
      "loss": 0.3824,
      "step": 181
    },
    {
      "epoch": 0.55,
      "learning_rate": 2e-05,
      "loss": 0.3689,
      "step": 182
    },
    {
      "epoch": 0.56,
      "learning_rate": 2e-05,
      "loss": 0.3719,
      "step": 183
    },
    {
      "epoch": 0.56,
      "learning_rate": 2e-05,
      "loss": 0.3568,
      "step": 184
    },
    {
      "epoch": 0.56,
      "learning_rate": 2e-05,
      "loss": 0.352,
      "step": 185
    },
    {
      "epoch": 0.57,
      "learning_rate": 2e-05,
      "loss": 0.3697,
      "step": 186
    },
    {
      "epoch": 0.57,
      "learning_rate": 2e-05,
      "loss": 0.3722,
      "step": 187
    },
    {
      "epoch": 0.57,
      "learning_rate": 2e-05,
      "loss": 0.37,
      "step": 188
    },
    {
      "epoch": 0.57,
      "learning_rate": 2e-05,
      "loss": 0.3687,
      "step": 189
    },
    {
      "epoch": 0.58,
      "learning_rate": 2e-05,
      "loss": 0.3494,
      "step": 190
    },
    {
      "epoch": 0.58,
      "learning_rate": 2e-05,
      "loss": 0.3527,
      "step": 191
    },
    {
      "epoch": 0.58,
      "learning_rate": 2e-05,
      "loss": 0.3724,
      "step": 192
    },
    {
      "epoch": 0.59,
      "learning_rate": 2e-05,
      "loss": 0.3589,
      "step": 193
    },
    {
      "epoch": 0.59,
      "learning_rate": 2e-05,
      "loss": 0.3661,
      "step": 194
    },
    {
      "epoch": 0.59,
      "learning_rate": 2e-05,
      "loss": 0.3733,
      "step": 195
    },
    {
      "epoch": 0.6,
      "learning_rate": 2e-05,
      "loss": 0.3699,
      "step": 196
    },
    {
      "epoch": 0.6,
      "learning_rate": 2e-05,
      "loss": 0.3585,
      "step": 197
    },
    {
      "epoch": 0.6,
      "learning_rate": 2e-05,
      "loss": 0.3535,
      "step": 198
    },
    {
      "epoch": 0.6,
      "learning_rate": 2e-05,
      "loss": 0.3599,
      "step": 199
    },
    {
      "epoch": 0.61,
      "learning_rate": 2e-05,
      "loss": 0.3673,
      "step": 200
    },
    {
      "epoch": 0.61,
      "learning_rate": 2e-05,
      "loss": 0.3583,
      "step": 201
    },
    {
      "epoch": 0.61,
      "learning_rate": 2e-05,
      "loss": 0.3646,
      "step": 202
    },
    {
      "epoch": 0.62,
      "learning_rate": 2e-05,
      "loss": 0.3533,
      "step": 203
    },
    {
      "epoch": 0.62,
      "learning_rate": 2e-05,
      "loss": 0.3515,
      "step": 204
    },
    {
      "epoch": 0.62,
      "learning_rate": 2e-05,
      "loss": 0.3671,
      "step": 205
    },
    {
      "epoch": 0.63,
      "learning_rate": 2e-05,
      "loss": 0.3443,
      "step": 206
    },
    {
      "epoch": 0.63,
      "learning_rate": 2e-05,
      "loss": 0.3478,
      "step": 207
    },
    {
      "epoch": 0.63,
      "learning_rate": 2e-05,
      "loss": 0.3682,
      "step": 208
    },
    {
      "epoch": 0.64,
      "learning_rate": 2e-05,
      "loss": 0.3736,
      "step": 209
    },
    {
      "epoch": 0.64,
      "learning_rate": 2e-05,
      "loss": 0.3583,
      "step": 210
    },
    {
      "epoch": 0.64,
      "learning_rate": 2e-05,
      "loss": 0.349,
      "step": 211
    },
    {
      "epoch": 0.64,
      "learning_rate": 2e-05,
      "loss": 0.3556,
      "step": 212
    },
    {
      "epoch": 0.65,
      "learning_rate": 2e-05,
      "loss": 0.3616,
      "step": 213
    },
    {
      "epoch": 0.65,
      "learning_rate": 2e-05,
      "loss": 0.3645,
      "step": 214
    },
    {
      "epoch": 0.65,
      "learning_rate": 2e-05,
      "loss": 0.3567,
      "step": 215
    },
    {
      "epoch": 0.66,
      "learning_rate": 2e-05,
      "loss": 0.3594,
      "step": 216
    },
    {
      "epoch": 0.66,
      "learning_rate": 2e-05,
      "loss": 0.3542,
      "step": 217
    },
    {
      "epoch": 0.66,
      "learning_rate": 2e-05,
      "loss": 0.3471,
      "step": 218
    },
    {
      "epoch": 0.67,
      "learning_rate": 2e-05,
      "loss": 0.3489,
      "step": 219
    },
    {
      "epoch": 0.67,
      "learning_rate": 2e-05,
      "loss": 0.3643,
      "step": 220
    },
    {
      "epoch": 0.67,
      "learning_rate": 2e-05,
      "loss": 0.3496,
      "step": 221
    },
    {
      "epoch": 0.67,
      "learning_rate": 2e-05,
      "loss": 0.3483,
      "step": 222
    },
    {
      "epoch": 0.68,
      "learning_rate": 2e-05,
      "loss": 0.3588,
      "step": 223
    },
    {
      "epoch": 0.68,
      "learning_rate": 2e-05,
      "loss": 0.3675,
      "step": 224
    },
    {
      "epoch": 0.68,
      "learning_rate": 2e-05,
      "loss": 0.3602,
      "step": 225
    },
    {
      "epoch": 0.69,
      "learning_rate": 2e-05,
      "loss": 0.3683,
      "step": 226
    },
    {
      "epoch": 0.69,
      "learning_rate": 2e-05,
      "loss": 0.3552,
      "step": 227
    },
    {
      "epoch": 0.69,
      "learning_rate": 2e-05,
      "loss": 0.3642,
      "step": 228
    },
    {
      "epoch": 0.7,
      "learning_rate": 2e-05,
      "loss": 0.3486,
      "step": 229
    },
    {
      "epoch": 0.7,
      "learning_rate": 2e-05,
      "loss": 0.3661,
      "step": 230
    },
    {
      "epoch": 0.7,
      "learning_rate": 2e-05,
      "loss": 0.3458,
      "step": 231
    },
    {
      "epoch": 0.71,
      "learning_rate": 2e-05,
      "loss": 0.3539,
      "step": 232
    },
    {
      "epoch": 0.71,
      "learning_rate": 2e-05,
      "loss": 0.3584,
      "step": 233
    },
    {
      "epoch": 0.71,
      "learning_rate": 2e-05,
      "loss": 0.3586,
      "step": 234
    },
    {
      "epoch": 0.71,
      "learning_rate": 2e-05,
      "loss": 0.3433,
      "step": 235
    },
    {
      "epoch": 0.72,
      "learning_rate": 2e-05,
      "loss": 0.3668,
      "step": 236
    },
    {
      "epoch": 0.72,
      "learning_rate": 2e-05,
      "loss": 0.3597,
      "step": 237
    },
    {
      "epoch": 0.72,
      "learning_rate": 2e-05,
      "loss": 0.3421,
      "step": 238
    },
    {
      "epoch": 0.73,
      "learning_rate": 2e-05,
      "loss": 0.3587,
      "step": 239
    },
    {
      "epoch": 0.73,
      "learning_rate": 2e-05,
      "loss": 0.3507,
      "step": 240
    },
    {
      "epoch": 0.73,
      "learning_rate": 2e-05,
      "loss": 0.3428,
      "step": 241
    },
    {
      "epoch": 0.74,
      "learning_rate": 2e-05,
      "loss": 0.3566,
      "step": 242
    },
    {
      "epoch": 0.74,
      "learning_rate": 2e-05,
      "loss": 0.3498,
      "step": 243
    },
    {
      "epoch": 0.74,
      "learning_rate": 2e-05,
      "loss": 0.346,
      "step": 244
    },
    {
      "epoch": 0.74,
      "learning_rate": 2e-05,
      "loss": 0.349,
      "step": 245
    },
    {
      "epoch": 0.75,
      "learning_rate": 2e-05,
      "loss": 0.355,
      "step": 246
    },
    {
      "epoch": 0.75,
      "learning_rate": 2e-05,
      "loss": 0.338,
      "step": 247
    },
    {
      "epoch": 0.75,
      "learning_rate": 2e-05,
      "loss": 0.3609,
      "step": 248
    },
    {
      "epoch": 0.76,
      "learning_rate": 2e-05,
      "loss": 0.3536,
      "step": 249
    },
    {
      "epoch": 0.76,
      "learning_rate": 2e-05,
      "loss": 0.3632,
      "step": 250
    },
    {
      "epoch": 0.76,
      "learning_rate": 2e-05,
      "loss": 0.3639,
      "step": 251
    },
    {
      "epoch": 0.77,
      "learning_rate": 2e-05,
      "loss": 0.3432,
      "step": 252
    },
    {
      "epoch": 0.77,
      "learning_rate": 2e-05,
      "loss": 0.3566,
      "step": 253
    },
    {
      "epoch": 0.77,
      "learning_rate": 2e-05,
      "loss": 0.3559,
      "step": 254
    },
    {
      "epoch": 0.78,
      "learning_rate": 2e-05,
      "loss": 0.3585,
      "step": 255
    },
    {
      "epoch": 0.78,
      "learning_rate": 2e-05,
      "loss": 0.3558,
      "step": 256
    },
    {
      "epoch": 0.78,
      "learning_rate": 2e-05,
      "loss": 0.3399,
      "step": 257
    },
    {
      "epoch": 0.78,
      "learning_rate": 2e-05,
      "loss": 0.3629,
      "step": 258
    },
    {
      "epoch": 0.79,
      "learning_rate": 2e-05,
      "loss": 0.3356,
      "step": 259
    },
    {
      "epoch": 0.79,
      "learning_rate": 2e-05,
      "loss": 0.3482,
      "step": 260
    },
    {
      "epoch": 0.79,
      "learning_rate": 2e-05,
      "loss": 0.3452,
      "step": 261
    },
    {
      "epoch": 0.8,
      "learning_rate": 2e-05,
      "loss": 0.3524,
      "step": 262
    },
    {
      "epoch": 0.8,
      "learning_rate": 2e-05,
      "loss": 0.3516,
      "step": 263
    },
    {
      "epoch": 0.8,
      "learning_rate": 2e-05,
      "loss": 0.3578,
      "step": 264
    },
    {
      "epoch": 0.81,
      "learning_rate": 2e-05,
      "loss": 0.3586,
      "step": 265
    },
    {
      "epoch": 0.81,
      "learning_rate": 2e-05,
      "loss": 0.3617,
      "step": 266
    },
    {
      "epoch": 0.81,
      "learning_rate": 2e-05,
      "loss": 0.3553,
      "step": 267
    },
    {
      "epoch": 0.81,
      "learning_rate": 2e-05,
      "loss": 0.3365,
      "step": 268
    },
    {
      "epoch": 0.82,
      "learning_rate": 2e-05,
      "loss": 0.3527,
      "step": 269
    },
    {
      "epoch": 0.82,
      "learning_rate": 2e-05,
      "loss": 0.3363,
      "step": 270
    },
    {
      "epoch": 0.82,
      "learning_rate": 2e-05,
      "loss": 0.3531,
      "step": 271
    },
    {
      "epoch": 0.83,
      "learning_rate": 2e-05,
      "loss": 0.3575,
      "step": 272
    },
    {
      "epoch": 0.83,
      "learning_rate": 2e-05,
      "loss": 0.3326,
      "step": 273
    },
    {
      "epoch": 0.83,
      "learning_rate": 2e-05,
      "loss": 0.363,
      "step": 274
    },
    {
      "epoch": 0.84,
      "learning_rate": 2e-05,
      "loss": 0.3463,
      "step": 275
    },
    {
      "epoch": 0.84,
      "learning_rate": 2e-05,
      "loss": 0.3504,
      "step": 276
    },
    {
      "epoch": 0.84,
      "learning_rate": 2e-05,
      "loss": 0.3554,
      "step": 277
    },
    {
      "epoch": 0.84,
      "learning_rate": 2e-05,
      "loss": 0.345,
      "step": 278
    },
    {
      "epoch": 0.85,
      "learning_rate": 2e-05,
      "loss": 0.3459,
      "step": 279
    },
    {
      "epoch": 0.85,
      "learning_rate": 2e-05,
      "loss": 0.3658,
      "step": 280
    },
    {
      "epoch": 0.85,
      "learning_rate": 2e-05,
      "loss": 0.3435,
      "step": 281
    },
    {
      "epoch": 0.86,
      "learning_rate": 2e-05,
      "loss": 0.3497,
      "step": 282
    },
    {
      "epoch": 0.86,
      "learning_rate": 2e-05,
      "loss": 0.3425,
      "step": 283
    },
    {
      "epoch": 0.86,
      "learning_rate": 2e-05,
      "loss": 0.3513,
      "step": 284
    },
    {
      "epoch": 0.87,
      "learning_rate": 2e-05,
      "loss": 0.3477,
      "step": 285
    },
    {
      "epoch": 0.87,
      "learning_rate": 2e-05,
      "loss": 0.3522,
      "step": 286
    },
    {
      "epoch": 0.87,
      "learning_rate": 2e-05,
      "loss": 0.3465,
      "step": 287
    },
    {
      "epoch": 0.88,
      "learning_rate": 2e-05,
      "loss": 0.3547,
      "step": 288
    },
    {
      "epoch": 0.88,
      "learning_rate": 2e-05,
      "loss": 0.3427,
      "step": 289
    },
    {
      "epoch": 0.88,
      "learning_rate": 2e-05,
      "loss": 0.3353,
      "step": 290
    },
    {
      "epoch": 0.88,
      "learning_rate": 2e-05,
      "loss": 0.3588,
      "step": 291
    },
    {
      "epoch": 0.89,
      "learning_rate": 2e-05,
      "loss": 0.3505,
      "step": 292
    },
    {
      "epoch": 0.89,
      "learning_rate": 2e-05,
      "loss": 0.3387,
      "step": 293
    },
    {
      "epoch": 0.89,
      "learning_rate": 2e-05,
      "loss": 0.3586,
      "step": 294
    },
    {
      "epoch": 0.9,
      "learning_rate": 2e-05,
      "loss": 0.3511,
      "step": 295
    },
    {
      "epoch": 0.9,
      "learning_rate": 2e-05,
      "loss": 0.3417,
      "step": 296
    },
    {
      "epoch": 0.9,
      "learning_rate": 2e-05,
      "loss": 0.3513,
      "step": 297
    },
    {
      "epoch": 0.91,
      "learning_rate": 2e-05,
      "loss": 0.3453,
      "step": 298
    },
    {
      "epoch": 0.91,
      "learning_rate": 2e-05,
      "loss": 0.3462,
      "step": 299
    },
    {
      "epoch": 0.91,
      "learning_rate": 2e-05,
      "loss": 0.3458,
      "step": 300
    },
    {
      "epoch": 0.91,
      "learning_rate": 2e-05,
      "loss": 0.3567,
      "step": 301
    },
    {
      "epoch": 0.92,
      "learning_rate": 2e-05,
      "loss": 0.3423,
      "step": 302
    },
    {
      "epoch": 0.92,
      "learning_rate": 2e-05,
      "loss": 0.3399,
      "step": 303
    },
    {
      "epoch": 0.92,
      "learning_rate": 2e-05,
      "loss": 0.3397,
      "step": 304
    },
    {
      "epoch": 0.93,
      "learning_rate": 2e-05,
      "loss": 0.3333,
      "step": 305
    },
    {
      "epoch": 0.93,
      "learning_rate": 2e-05,
      "loss": 0.3397,
      "step": 306
    },
    {
      "epoch": 0.93,
      "learning_rate": 2e-05,
      "loss": 0.3372,
      "step": 307
    },
    {
      "epoch": 0.94,
      "learning_rate": 2e-05,
      "loss": 0.3422,
      "step": 308
    },
    {
      "epoch": 0.94,
      "learning_rate": 2e-05,
      "loss": 0.337,
      "step": 309
    },
    {
      "epoch": 0.94,
      "learning_rate": 2e-05,
      "loss": 0.3505,
      "step": 310
    },
    {
      "epoch": 0.95,
      "learning_rate": 2e-05,
      "loss": 0.3428,
      "step": 311
    },
    {
      "epoch": 0.95,
      "learning_rate": 2e-05,
      "loss": 0.3392,
      "step": 312
    },
    {
      "epoch": 0.95,
      "learning_rate": 2e-05,
      "loss": 0.3409,
      "step": 313
    },
    {
      "epoch": 0.95,
      "learning_rate": 2e-05,
      "loss": 0.3426,
      "step": 314
    },
    {
      "epoch": 0.96,
      "learning_rate": 2e-05,
      "loss": 0.3297,
      "step": 315
    },
    {
      "epoch": 0.96,
      "learning_rate": 2e-05,
      "loss": 0.3408,
      "step": 316
    },
    {
      "epoch": 0.96,
      "learning_rate": 2e-05,
      "loss": 0.3488,
      "step": 317
    },
    {
      "epoch": 0.97,
      "learning_rate": 2e-05,
      "loss": 0.357,
      "step": 318
    },
    {
      "epoch": 0.97,
      "learning_rate": 2e-05,
      "loss": 0.3368,
      "step": 319
    },
    {
      "epoch": 0.97,
      "learning_rate": 2e-05,
      "loss": 0.3354,
      "step": 320
    },
    {
      "epoch": 0.98,
      "learning_rate": 2e-05,
      "loss": 0.3407,
      "step": 321
    },
    {
      "epoch": 0.98,
      "learning_rate": 2e-05,
      "loss": 0.3345,
      "step": 322
    },
    {
      "epoch": 0.98,
      "learning_rate": 2e-05,
      "loss": 0.3358,
      "step": 323
    },
    {
      "epoch": 0.98,
      "learning_rate": 2e-05,
      "loss": 0.3383,
      "step": 324
    },
    {
      "epoch": 0.99,
      "learning_rate": 2e-05,
      "loss": 0.3373,
      "step": 325
    },
    {
      "epoch": 0.99,
      "learning_rate": 2e-05,
      "loss": 0.3382,
      "step": 326
    },
    {
      "epoch": 0.99,
      "learning_rate": 2e-05,
      "loss": 0.3357,
      "step": 327
    },
    {
      "epoch": 1.0,
      "learning_rate": 2e-05,
      "loss": 0.3463,
      "step": 328
    },
    {
      "epoch": 1.0,
      "learning_rate": 2e-05,
      "loss": 0.3467,
      "step": 329
    },
    {
      "epoch": 1.0,
      "learning_rate": 2e-05,
      "loss": 0.3142,
      "step": 330
    },
    {
      "epoch": 1.01,
      "learning_rate": 2e-05,
      "loss": 0.315,
      "step": 331
    },
    {
      "epoch": 1.01,
      "learning_rate": 2e-05,
      "loss": 0.3053,
      "step": 332
    },
    {
      "epoch": 1.01,
      "learning_rate": 2e-05,
      "loss": 0.3081,
      "step": 333
    },
    {
      "epoch": 1.02,
      "learning_rate": 2e-05,
      "loss": 0.314,
      "step": 334
    },
    {
      "epoch": 1.02,
      "learning_rate": 2e-05,
      "loss": 0.2991,
      "step": 335
    },
    {
      "epoch": 1.02,
      "learning_rate": 2e-05,
      "loss": 0.302,
      "step": 336
    },
    {
      "epoch": 1.02,
      "learning_rate": 2e-05,
      "loss": 0.3072,
      "step": 337
    },
    {
      "epoch": 1.03,
      "learning_rate": 2e-05,
      "loss": 0.3143,
      "step": 338
    },
    {
      "epoch": 1.03,
      "learning_rate": 2e-05,
      "loss": 0.3127,
      "step": 339
    },
    {
      "epoch": 1.03,
      "learning_rate": 2e-05,
      "loss": 0.3098,
      "step": 340
    },
    {
      "epoch": 1.04,
      "learning_rate": 2e-05,
      "loss": 0.3176,
      "step": 341
    },
    {
      "epoch": 1.04,
      "learning_rate": 2e-05,
      "loss": 0.306,
      "step": 342
    },
    {
      "epoch": 1.04,
      "learning_rate": 2e-05,
      "loss": 0.3155,
      "step": 343
    },
    {
      "epoch": 1.05,
      "learning_rate": 2e-05,
      "loss": 0.3074,
      "step": 344
    },
    {
      "epoch": 1.05,
      "learning_rate": 2e-05,
      "loss": 0.3091,
      "step": 345
    },
    {
      "epoch": 1.05,
      "learning_rate": 2e-05,
      "loss": 0.3042,
      "step": 346
    },
    {
      "epoch": 1.05,
      "learning_rate": 2e-05,
      "loss": 0.3012,
      "step": 347
    },
    {
      "epoch": 1.06,
      "learning_rate": 2e-05,
      "loss": 0.3083,
      "step": 348
    },
    {
      "epoch": 1.06,
      "learning_rate": 2e-05,
      "loss": 0.3099,
      "step": 349
    },
    {
      "epoch": 1.06,
      "learning_rate": 2e-05,
      "loss": 0.3072,
      "step": 350
    },
    {
      "epoch": 1.07,
      "learning_rate": 2e-05,
      "loss": 0.3024,
      "step": 351
    },
    {
      "epoch": 1.07,
      "learning_rate": 2e-05,
      "loss": 0.3024,
      "step": 352
    },
    {
      "epoch": 1.07,
      "learning_rate": 2e-05,
      "loss": 0.3051,
      "step": 353
    },
    {
      "epoch": 1.08,
      "learning_rate": 2e-05,
      "loss": 0.3021,
      "step": 354
    },
    {
      "epoch": 1.08,
      "learning_rate": 2e-05,
      "loss": 0.3039,
      "step": 355
    },
    {
      "epoch": 1.08,
      "learning_rate": 2e-05,
      "loss": 0.3036,
      "step": 356
    },
    {
      "epoch": 1.09,
      "learning_rate": 2e-05,
      "loss": 0.32,
      "step": 357
    },
    {
      "epoch": 1.09,
      "learning_rate": 2e-05,
      "loss": 0.3147,
      "step": 358
    },
    {
      "epoch": 1.09,
      "learning_rate": 2e-05,
      "loss": 0.3095,
      "step": 359
    },
    {
      "epoch": 1.09,
      "learning_rate": 2e-05,
      "loss": 0.3196,
      "step": 360
    },
    {
      "epoch": 1.1,
      "learning_rate": 2e-05,
      "loss": 0.3205,
      "step": 361
    },
    {
      "epoch": 1.1,
      "learning_rate": 2e-05,
      "loss": 0.3067,
      "step": 362
    },
    {
      "epoch": 1.1,
      "learning_rate": 2e-05,
      "loss": 0.3028,
      "step": 363
    },
    {
      "epoch": 1.11,
      "learning_rate": 2e-05,
      "loss": 0.3053,
      "step": 364
    },
    {
      "epoch": 1.11,
      "learning_rate": 2e-05,
      "loss": 0.306,
      "step": 365
    },
    {
      "epoch": 1.11,
      "learning_rate": 2e-05,
      "loss": 0.3123,
      "step": 366
    },
    {
      "epoch": 1.12,
      "learning_rate": 2e-05,
      "loss": 0.308,
      "step": 367
    },
    {
      "epoch": 1.12,
      "learning_rate": 2e-05,
      "loss": 0.3101,
      "step": 368
    },
    {
      "epoch": 1.12,
      "learning_rate": 2e-05,
      "loss": 0.3038,
      "step": 369
    },
    {
      "epoch": 1.12,
      "learning_rate": 2e-05,
      "loss": 0.311,
      "step": 370
    },
    {
      "epoch": 1.13,
      "learning_rate": 2e-05,
      "loss": 0.3174,
      "step": 371
    },
    {
      "epoch": 1.13,
      "learning_rate": 2e-05,
      "loss": 0.3138,
      "step": 372
    },
    {
      "epoch": 1.13,
      "learning_rate": 2e-05,
      "loss": 0.3082,
      "step": 373
    },
    {
      "epoch": 1.14,
      "learning_rate": 2e-05,
      "loss": 0.3073,
      "step": 374
    },
    {
      "epoch": 1.14,
      "learning_rate": 2e-05,
      "loss": 0.3026,
      "step": 375
    },
    {
      "epoch": 1.14,
      "learning_rate": 2e-05,
      "loss": 0.3062,
      "step": 376
    },
    {
      "epoch": 1.15,
      "learning_rate": 2e-05,
      "loss": 0.3238,
      "step": 377
    },
    {
      "epoch": 1.15,
      "learning_rate": 2e-05,
      "loss": 0.3058,
      "step": 378
    },
    {
      "epoch": 1.15,
      "learning_rate": 2e-05,
      "loss": 0.3179,
      "step": 379
    },
    {
      "epoch": 1.16,
      "learning_rate": 2e-05,
      "loss": 0.3097,
      "step": 380
    },
    {
      "epoch": 1.16,
      "learning_rate": 2e-05,
      "loss": 0.3045,
      "step": 381
    },
    {
      "epoch": 1.16,
      "learning_rate": 2e-05,
      "loss": 0.3095,
      "step": 382
    },
    {
      "epoch": 1.16,
      "learning_rate": 2e-05,
      "loss": 0.301,
      "step": 383
    },
    {
      "epoch": 1.17,
      "learning_rate": 2e-05,
      "loss": 0.3121,
      "step": 384
    },
    {
      "epoch": 1.17,
      "learning_rate": 2e-05,
      "loss": 0.2999,
      "step": 385
    },
    {
      "epoch": 1.17,
      "learning_rate": 2e-05,
      "loss": 0.3049,
      "step": 386
    },
    {
      "epoch": 1.18,
      "learning_rate": 2e-05,
      "loss": 0.3169,
      "step": 387
    },
    {
      "epoch": 1.18,
      "learning_rate": 2e-05,
      "loss": 0.3008,
      "step": 388
    },
    {
      "epoch": 1.18,
      "learning_rate": 2e-05,
      "loss": 0.318,
      "step": 389
    },
    {
      "epoch": 1.19,
      "learning_rate": 2e-05,
      "loss": 0.3069,
      "step": 390
    },
    {
      "epoch": 1.19,
      "learning_rate": 2e-05,
      "loss": 0.3092,
      "step": 391
    },
    {
      "epoch": 1.19,
      "learning_rate": 2e-05,
      "loss": 0.3088,
      "step": 392
    },
    {
      "epoch": 1.19,
      "learning_rate": 2e-05,
      "loss": 0.2973,
      "step": 393
    },
    {
      "epoch": 1.2,
      "learning_rate": 2e-05,
      "loss": 0.3141,
      "step": 394
    },
    {
      "epoch": 1.2,
      "learning_rate": 2e-05,
      "loss": 0.3054,
      "step": 395
    },
    {
      "epoch": 1.2,
      "learning_rate": 2e-05,
      "loss": 0.3135,
      "step": 396
    },
    {
      "epoch": 1.21,
      "learning_rate": 2e-05,
      "loss": 0.2988,
      "step": 397
    },
    {
      "epoch": 1.21,
      "learning_rate": 2e-05,
      "loss": 0.2926,
      "step": 398
    },
    {
      "epoch": 1.21,
      "learning_rate": 2e-05,
      "loss": 0.3081,
      "step": 399
    },
    {
      "epoch": 1.22,
      "learning_rate": 2e-05,
      "loss": 0.2924,
      "step": 400
    },
    {
      "epoch": 1.22,
      "learning_rate": 2e-05,
      "loss": 0.3077,
      "step": 401
    },
    {
      "epoch": 1.22,
      "learning_rate": 2e-05,
      "loss": 0.318,
      "step": 402
    },
    {
      "epoch": 1.22,
      "learning_rate": 2e-05,
      "loss": 0.303,
      "step": 403
    },
    {
      "epoch": 1.23,
      "learning_rate": 2e-05,
      "loss": 0.308,
      "step": 404
    },
    {
      "epoch": 1.23,
      "learning_rate": 2e-05,
      "loss": 0.3024,
      "step": 405
    },
    {
      "epoch": 1.23,
      "learning_rate": 2e-05,
      "loss": 0.2927,
      "step": 406
    },
    {
      "epoch": 1.24,
      "learning_rate": 2e-05,
      "loss": 0.306,
      "step": 407
    },
    {
      "epoch": 1.24,
      "learning_rate": 2e-05,
      "loss": 0.3026,
      "step": 408
    },
    {
      "epoch": 1.24,
      "learning_rate": 2e-05,
      "loss": 0.3042,
      "step": 409
    },
    {
      "epoch": 1.25,
      "learning_rate": 2e-05,
      "loss": 0.315,
      "step": 410
    },
    {
      "epoch": 1.25,
      "learning_rate": 2e-05,
      "loss": 0.2996,
      "step": 411
    },
    {
      "epoch": 1.25,
      "learning_rate": 2e-05,
      "loss": 0.3127,
      "step": 412
    },
    {
      "epoch": 1.26,
      "learning_rate": 2e-05,
      "loss": 0.2998,
      "step": 413
    },
    {
      "epoch": 1.26,
      "learning_rate": 2e-05,
      "loss": 0.3049,
      "step": 414
    },
    {
      "epoch": 1.26,
      "learning_rate": 2e-05,
      "loss": 0.3074,
      "step": 415
    },
    {
      "epoch": 1.26,
      "learning_rate": 2e-05,
      "loss": 0.3008,
      "step": 416
    },
    {
      "epoch": 1.27,
      "learning_rate": 2e-05,
      "loss": 0.308,
      "step": 417
    },
    {
      "epoch": 1.27,
      "learning_rate": 2e-05,
      "loss": 0.3013,
      "step": 418
    },
    {
      "epoch": 1.27,
      "learning_rate": 2e-05,
      "loss": 0.3018,
      "step": 419
    },
    {
      "epoch": 1.28,
      "learning_rate": 2e-05,
      "loss": 0.3106,
      "step": 420
    },
    {
      "epoch": 1.28,
      "learning_rate": 2e-05,
      "loss": 0.2975,
      "step": 421
    },
    {
      "epoch": 1.28,
      "learning_rate": 2e-05,
      "loss": 0.3003,
      "step": 422
    },
    {
      "epoch": 1.29,
      "learning_rate": 2e-05,
      "loss": 0.2911,
      "step": 423
    },
    {
      "epoch": 1.29,
      "learning_rate": 2e-05,
      "loss": 0.3087,
      "step": 424
    },
    {
      "epoch": 1.29,
      "learning_rate": 2e-05,
      "loss": 0.2979,
      "step": 425
    },
    {
      "epoch": 1.29,
      "learning_rate": 2e-05,
      "loss": 0.2995,
      "step": 426
    },
    {
      "epoch": 1.3,
      "learning_rate": 2e-05,
      "loss": 0.3073,
      "step": 427
    },
    {
      "epoch": 1.3,
      "learning_rate": 2e-05,
      "loss": 0.3116,
      "step": 428
    },
    {
      "epoch": 1.3,
      "learning_rate": 2e-05,
      "loss": 0.305,
      "step": 429
    },
    {
      "epoch": 1.31,
      "learning_rate": 2e-05,
      "loss": 0.2969,
      "step": 430
    },
    {
      "epoch": 1.31,
      "learning_rate": 2e-05,
      "loss": 0.3169,
      "step": 431
    },
    {
      "epoch": 1.31,
      "learning_rate": 2e-05,
      "loss": 0.3095,
      "step": 432
    },
    {
      "epoch": 1.32,
      "learning_rate": 2e-05,
      "loss": 0.3053,
      "step": 433
    },
    {
      "epoch": 1.32,
      "learning_rate": 2e-05,
      "loss": 0.3058,
      "step": 434
    },
    {
      "epoch": 1.32,
      "learning_rate": 2e-05,
      "loss": 0.3092,
      "step": 435
    },
    {
      "epoch": 1.33,
      "learning_rate": 2e-05,
      "loss": 0.3165,
      "step": 436
    },
    {
      "epoch": 1.33,
      "learning_rate": 2e-05,
      "loss": 0.3122,
      "step": 437
    },
    {
      "epoch": 1.33,
      "learning_rate": 2e-05,
      "loss": 0.3114,
      "step": 438
    },
    {
      "epoch": 1.33,
      "learning_rate": 2e-05,
      "loss": 0.302,
      "step": 439
    },
    {
      "epoch": 1.34,
      "learning_rate": 2e-05,
      "loss": 0.2992,
      "step": 440
    },
    {
      "epoch": 1.34,
      "learning_rate": 2e-05,
      "loss": 0.3033,
      "step": 441
    },
    {
      "epoch": 1.34,
      "learning_rate": 2e-05,
      "loss": 0.3053,
      "step": 442
    },
    {
      "epoch": 1.35,
      "learning_rate": 2e-05,
      "loss": 0.3106,
      "step": 443
    },
    {
      "epoch": 1.35,
      "learning_rate": 2e-05,
      "loss": 0.3121,
      "step": 444
    },
    {
      "epoch": 1.35,
      "learning_rate": 2e-05,
      "loss": 0.3049,
      "step": 445
    },
    {
      "epoch": 1.36,
      "learning_rate": 2e-05,
      "loss": 0.303,
      "step": 446
    },
    {
      "epoch": 1.36,
      "learning_rate": 2e-05,
      "loss": 0.3034,
      "step": 447
    },
    {
      "epoch": 1.36,
      "learning_rate": 2e-05,
      "loss": 0.306,
      "step": 448
    },
    {
      "epoch": 1.36,
      "learning_rate": 2e-05,
      "loss": 0.3194,
      "step": 449
    },
    {
      "epoch": 1.37,
      "learning_rate": 2e-05,
      "loss": 0.3021,
      "step": 450
    },
    {
      "epoch": 1.37,
      "learning_rate": 2e-05,
      "loss": 0.3145,
      "step": 451
    },
    {
      "epoch": 1.37,
      "learning_rate": 2e-05,
      "loss": 0.2913,
      "step": 452
    },
    {
      "epoch": 1.38,
      "learning_rate": 2e-05,
      "loss": 0.2972,
      "step": 453
    },
    {
      "epoch": 1.38,
      "learning_rate": 2e-05,
      "loss": 0.3137,
      "step": 454
    },
    {
      "epoch": 1.38,
      "learning_rate": 2e-05,
      "loss": 0.3043,
      "step": 455
    },
    {
      "epoch": 1.39,
      "learning_rate": 2e-05,
      "loss": 0.312,
      "step": 456
    },
    {
      "epoch": 1.39,
      "learning_rate": 2e-05,
      "loss": 0.3024,
      "step": 457
    },
    {
      "epoch": 1.39,
      "learning_rate": 2e-05,
      "loss": 0.298,
      "step": 458
    },
    {
      "epoch": 1.4,
      "learning_rate": 2e-05,
      "loss": 0.3053,
      "step": 459
    },
    {
      "epoch": 1.4,
      "learning_rate": 2e-05,
      "loss": 0.3003,
      "step": 460
    },
    {
      "epoch": 1.4,
      "learning_rate": 2e-05,
      "loss": 0.3058,
      "step": 461
    },
    {
      "epoch": 1.4,
      "learning_rate": 2e-05,
      "loss": 0.3108,
      "step": 462
    },
    {
      "epoch": 1.41,
      "learning_rate": 2e-05,
      "loss": 0.3071,
      "step": 463
    },
    {
      "epoch": 1.41,
      "learning_rate": 2e-05,
      "loss": 0.3085,
      "step": 464
    },
    {
      "epoch": 1.41,
      "learning_rate": 2e-05,
      "loss": 0.3193,
      "step": 465
    },
    {
      "epoch": 1.42,
      "learning_rate": 2e-05,
      "loss": 0.3071,
      "step": 466
    },
    {
      "epoch": 1.42,
      "learning_rate": 2e-05,
      "loss": 0.3024,
      "step": 467
    },
    {
      "epoch": 1.42,
      "learning_rate": 2e-05,
      "loss": 0.3034,
      "step": 468
    },
    {
      "epoch": 1.43,
      "learning_rate": 2e-05,
      "loss": 0.3034,
      "step": 469
    },
    {
      "epoch": 1.43,
      "learning_rate": 2e-05,
      "loss": 0.2955,
      "step": 470
    },
    {
      "epoch": 1.43,
      "learning_rate": 2e-05,
      "loss": 0.3028,
      "step": 471
    },
    {
      "epoch": 1.43,
      "learning_rate": 2e-05,
      "loss": 0.2976,
      "step": 472
    },
    {
      "epoch": 1.44,
      "learning_rate": 2e-05,
      "loss": 0.3071,
      "step": 473
    },
    {
      "epoch": 1.44,
      "learning_rate": 2e-05,
      "loss": 0.3066,
      "step": 474
    },
    {
      "epoch": 1.44,
      "learning_rate": 2e-05,
      "loss": 0.3037,
      "step": 475
    },
    {
      "epoch": 1.45,
      "learning_rate": 2e-05,
      "loss": 0.298,
      "step": 476
    },
    {
      "epoch": 1.45,
      "learning_rate": 2e-05,
      "loss": 0.3107,
      "step": 477
    },
    {
      "epoch": 1.45,
      "learning_rate": 2e-05,
      "loss": 0.2898,
      "step": 478
    },
    {
      "epoch": 1.46,
      "learning_rate": 2e-05,
      "loss": 0.305,
      "step": 479
    },
    {
      "epoch": 1.46,
      "learning_rate": 2e-05,
      "loss": 0.3021,
      "step": 480
    },
    {
      "epoch": 1.46,
      "learning_rate": 2e-05,
      "loss": 0.3018,
      "step": 481
    },
    {
      "epoch": 1.47,
      "learning_rate": 2e-05,
      "loss": 0.3071,
      "step": 482
    },
    {
      "epoch": 1.47,
      "learning_rate": 2e-05,
      "loss": 0.2935,
      "step": 483
    },
    {
      "epoch": 1.47,
      "learning_rate": 2e-05,
      "loss": 0.3223,
      "step": 484
    },
    {
      "epoch": 1.47,
      "learning_rate": 2e-05,
      "loss": 0.2929,
      "step": 485
    },
    {
      "epoch": 1.48,
      "learning_rate": 2e-05,
      "loss": 0.29,
      "step": 486
    },
    {
      "epoch": 1.48,
      "learning_rate": 2e-05,
      "loss": 0.3063,
      "step": 487
    },
    {
      "epoch": 1.48,
      "learning_rate": 2e-05,
      "loss": 0.3101,
      "step": 488
    },
    {
      "epoch": 1.49,
      "learning_rate": 2e-05,
      "loss": 0.3103,
      "step": 489
    },
    {
      "epoch": 1.49,
      "learning_rate": 2e-05,
      "loss": 0.2953,
      "step": 490
    },
    {
      "epoch": 1.49,
      "learning_rate": 2e-05,
      "loss": 0.2966,
      "step": 491
    },
    {
      "epoch": 1.5,
      "learning_rate": 2e-05,
      "loss": 0.3041,
      "step": 492
    },
    {
      "epoch": 1.5,
      "learning_rate": 2e-05,
      "loss": 0.2983,
      "step": 493
    },
    {
      "epoch": 1.5,
      "learning_rate": 2e-05,
      "loss": 0.2886,
      "step": 494
    },
    {
      "epoch": 1.5,
      "learning_rate": 2e-05,
      "loss": 0.2905,
      "step": 495
    },
    {
      "epoch": 1.51,
      "learning_rate": 2e-05,
      "loss": 0.3069,
      "step": 496
    },
    {
      "epoch": 1.51,
      "learning_rate": 2e-05,
      "loss": 0.3085,
      "step": 497
    },
    {
      "epoch": 1.51,
      "learning_rate": 2e-05,
      "loss": 0.2962,
      "step": 498
    },
    {
      "epoch": 1.52,
      "learning_rate": 2e-05,
      "loss": 0.3066,
      "step": 499
    },
    {
      "epoch": 1.52,
      "learning_rate": 2e-05,
      "loss": 0.2906,
      "step": 500
    },
    {
      "epoch": 1.52,
      "learning_rate": 2e-05,
      "loss": 0.308,
      "step": 501
    },
    {
      "epoch": 1.53,
      "learning_rate": 2e-05,
      "loss": 0.2948,
      "step": 502
    },
    {
      "epoch": 1.53,
      "learning_rate": 2e-05,
      "loss": 0.3045,
      "step": 503
    },
    {
      "epoch": 1.53,
      "learning_rate": 2e-05,
      "loss": 0.2992,
      "step": 504
    },
    {
      "epoch": 1.53,
      "learning_rate": 2e-05,
      "loss": 0.3058,
      "step": 505
    },
    {
      "epoch": 1.54,
      "learning_rate": 2e-05,
      "loss": 0.2926,
      "step": 506
    },
    {
      "epoch": 1.54,
      "learning_rate": 2e-05,
      "loss": 0.294,
      "step": 507
    },
    {
      "epoch": 1.54,
      "learning_rate": 2e-05,
      "loss": 0.3112,
      "step": 508
    },
    {
      "epoch": 1.55,
      "learning_rate": 2e-05,
      "loss": 0.3009,
      "step": 509
    },
    {
      "epoch": 1.55,
      "learning_rate": 2e-05,
      "loss": 0.2955,
      "step": 510
    },
    {
      "epoch": 1.55,
      "learning_rate": 2e-05,
      "loss": 0.302,
      "step": 511
    },
    {
      "epoch": 1.56,
      "learning_rate": 2e-05,
      "loss": 0.3025,
      "step": 512
    },
    {
      "epoch": 1.56,
      "learning_rate": 2e-05,
      "loss": 0.3042,
      "step": 513
    },
    {
      "epoch": 1.56,
      "learning_rate": 2e-05,
      "loss": 0.3028,
      "step": 514
    },
    {
      "epoch": 1.57,
      "learning_rate": 2e-05,
      "loss": 0.3024,
      "step": 515
    },
    {
      "epoch": 1.57,
      "learning_rate": 2e-05,
      "loss": 0.2989,
      "step": 516
    },
    {
      "epoch": 1.57,
      "learning_rate": 2e-05,
      "loss": 0.3053,
      "step": 517
    },
    {
      "epoch": 1.57,
      "learning_rate": 2e-05,
      "loss": 0.3015,
      "step": 518
    },
    {
      "epoch": 1.58,
      "learning_rate": 2e-05,
      "loss": 0.3032,
      "step": 519
    },
    {
      "epoch": 1.58,
      "learning_rate": 2e-05,
      "loss": 0.2899,
      "step": 520
    },
    {
      "epoch": 1.58,
      "learning_rate": 2e-05,
      "loss": 0.3128,
      "step": 521
    },
    {
      "epoch": 1.59,
      "learning_rate": 2e-05,
      "loss": 0.3009,
      "step": 522
    },
    {
      "epoch": 1.59,
      "learning_rate": 2e-05,
      "loss": 0.2904,
      "step": 523
    },
    {
      "epoch": 1.59,
      "learning_rate": 2e-05,
      "loss": 0.3001,
      "step": 524
    },
    {
      "epoch": 1.6,
      "learning_rate": 2e-05,
      "loss": 0.2978,
      "step": 525
    },
    {
      "epoch": 1.6,
      "learning_rate": 2e-05,
      "loss": 0.2967,
      "step": 526
    },
    {
      "epoch": 1.6,
      "learning_rate": 2e-05,
      "loss": 0.2927,
      "step": 527
    },
    {
      "epoch": 1.6,
      "learning_rate": 2e-05,
      "loss": 0.3053,
      "step": 528
    },
    {
      "epoch": 1.61,
      "learning_rate": 2e-05,
      "loss": 0.297,
      "step": 529
    },
    {
      "epoch": 1.61,
      "learning_rate": 2e-05,
      "loss": 0.3131,
      "step": 530
    },
    {
      "epoch": 1.61,
      "learning_rate": 2e-05,
      "loss": 0.2902,
      "step": 531
    },
    {
      "epoch": 1.62,
      "learning_rate": 2e-05,
      "loss": 0.2935,
      "step": 532
    },
    {
      "epoch": 1.62,
      "learning_rate": 2e-05,
      "loss": 0.3019,
      "step": 533
    },
    {
      "epoch": 1.62,
      "learning_rate": 2e-05,
      "loss": 0.2935,
      "step": 534
    },
    {
      "epoch": 1.63,
      "learning_rate": 2e-05,
      "loss": 0.2957,
      "step": 535
    },
    {
      "epoch": 1.63,
      "learning_rate": 2e-05,
      "loss": 0.3084,
      "step": 536
    },
    {
      "epoch": 1.63,
      "learning_rate": 2e-05,
      "loss": 0.3022,
      "step": 537
    },
    {
      "epoch": 1.64,
      "learning_rate": 2e-05,
      "loss": 0.3055,
      "step": 538
    },
    {
      "epoch": 1.64,
      "learning_rate": 2e-05,
      "loss": 0.3069,
      "step": 539
    },
    {
      "epoch": 1.64,
      "learning_rate": 2e-05,
      "loss": 0.2998,
      "step": 540
    },
    {
      "epoch": 1.64,
      "learning_rate": 2e-05,
      "loss": 0.3118,
      "step": 541
    },
    {
      "epoch": 1.65,
      "learning_rate": 2e-05,
      "loss": 0.3001,
      "step": 542
    },
    {
      "epoch": 1.65,
      "learning_rate": 2e-05,
      "loss": 0.3042,
      "step": 543
    },
    {
      "epoch": 1.65,
      "learning_rate": 2e-05,
      "loss": 0.2982,
      "step": 544
    },
    {
      "epoch": 1.66,
      "learning_rate": 2e-05,
      "loss": 0.2866,
      "step": 545
    },
    {
      "epoch": 1.66,
      "learning_rate": 2e-05,
      "loss": 0.298,
      "step": 546
    },
    {
      "epoch": 1.66,
      "learning_rate": 2e-05,
      "loss": 0.2918,
      "step": 547
    },
    {
      "epoch": 1.67,
      "learning_rate": 2e-05,
      "loss": 0.2982,
      "step": 548
    },
    {
      "epoch": 1.67,
      "learning_rate": 2e-05,
      "loss": 0.2971,
      "step": 549
    },
    {
      "epoch": 1.67,
      "learning_rate": 2e-05,
      "loss": 0.2864,
      "step": 550
    },
    {
      "epoch": 1.67,
      "learning_rate": 2e-05,
      "loss": 0.3034,
      "step": 551
    },
    {
      "epoch": 1.68,
      "learning_rate": 2e-05,
      "loss": 0.3005,
      "step": 552
    },
    {
      "epoch": 1.68,
      "learning_rate": 2e-05,
      "loss": 0.2991,
      "step": 553
    },
    {
      "epoch": 1.68,
      "learning_rate": 2e-05,
      "loss": 0.2961,
      "step": 554
    },
    {
      "epoch": 1.69,
      "learning_rate": 2e-05,
      "loss": 0.3022,
      "step": 555
    },
    {
      "epoch": 1.69,
      "learning_rate": 2e-05,
      "loss": 0.3021,
      "step": 556
    },
    {
      "epoch": 1.69,
      "learning_rate": 2e-05,
      "loss": 0.294,
      "step": 557
    },
    {
      "epoch": 1.7,
      "learning_rate": 2e-05,
      "loss": 0.2855,
      "step": 558
    },
    {
      "epoch": 1.7,
      "learning_rate": 2e-05,
      "loss": 0.3035,
      "step": 559
    },
    {
      "epoch": 1.7,
      "learning_rate": 2e-05,
      "loss": 0.2956,
      "step": 560
    },
    {
      "epoch": 1.71,
      "learning_rate": 2e-05,
      "loss": 0.3006,
      "step": 561
    },
    {
      "epoch": 1.71,
      "learning_rate": 2e-05,
      "loss": 0.294,
      "step": 562
    },
    {
      "epoch": 1.71,
      "learning_rate": 2e-05,
      "loss": 0.294,
      "step": 563
    },
    {
      "epoch": 1.71,
      "learning_rate": 2e-05,
      "loss": 0.296,
      "step": 564
    },
    {
      "epoch": 1.72,
      "learning_rate": 2e-05,
      "loss": 0.2844,
      "step": 565
    },
    {
      "epoch": 1.72,
      "learning_rate": 2e-05,
      "loss": 0.2923,
      "step": 566
    },
    {
      "epoch": 1.72,
      "learning_rate": 2e-05,
      "loss": 0.2866,
      "step": 567
    },
    {
      "epoch": 1.73,
      "learning_rate": 2e-05,
      "loss": 0.295,
      "step": 568
    },
    {
      "epoch": 1.73,
      "learning_rate": 2e-05,
      "loss": 0.2996,
      "step": 569
    },
    {
      "epoch": 1.73,
      "learning_rate": 2e-05,
      "loss": 0.2882,
      "step": 570
    },
    {
      "epoch": 1.74,
      "learning_rate": 2e-05,
      "loss": 0.2953,
      "step": 571
    },
    {
      "epoch": 1.74,
      "learning_rate": 2e-05,
      "loss": 0.2917,
      "step": 572
    },
    {
      "epoch": 1.74,
      "learning_rate": 2e-05,
      "loss": 0.2894,
      "step": 573
    },
    {
      "epoch": 1.74,
      "learning_rate": 2e-05,
      "loss": 0.2988,
      "step": 574
    },
    {
      "epoch": 1.75,
      "learning_rate": 2e-05,
      "loss": 0.2965,
      "step": 575
    },
    {
      "epoch": 1.75,
      "learning_rate": 2e-05,
      "loss": 0.2979,
      "step": 576
    },
    {
      "epoch": 1.75,
      "learning_rate": 2e-05,
      "loss": 0.3024,
      "step": 577
    },
    {
      "epoch": 1.76,
      "learning_rate": 2e-05,
      "loss": 0.3027,
      "step": 578
    },
    {
      "epoch": 1.76,
      "learning_rate": 2e-05,
      "loss": 0.293,
      "step": 579
    },
    {
      "epoch": 1.76,
      "learning_rate": 2e-05,
      "loss": 0.2873,
      "step": 580
    },
    {
      "epoch": 1.77,
      "learning_rate": 2e-05,
      "loss": 0.2861,
      "step": 581
    },
    {
      "epoch": 1.77,
      "learning_rate": 2e-05,
      "loss": 0.2999,
      "step": 582
    },
    {
      "epoch": 1.77,
      "learning_rate": 2e-05,
      "loss": 0.3011,
      "step": 583
    },
    {
      "epoch": 1.78,
      "learning_rate": 2e-05,
      "loss": 0.2855,
      "step": 584
    },
    {
      "epoch": 1.78,
      "learning_rate": 2e-05,
      "loss": 0.2927,
      "step": 585
    },
    {
      "epoch": 1.78,
      "learning_rate": 2e-05,
      "loss": 0.2973,
      "step": 586
    },
    {
      "epoch": 1.78,
      "learning_rate": 2e-05,
      "loss": 0.2963,
      "step": 587
    },
    {
      "epoch": 1.79,
      "learning_rate": 2e-05,
      "loss": 0.2935,
      "step": 588
    },
    {
      "epoch": 1.79,
      "learning_rate": 2e-05,
      "loss": 0.285,
      "step": 589
    },
    {
      "epoch": 1.79,
      "learning_rate": 2e-05,
      "loss": 0.2996,
      "step": 590
    },
    {
      "epoch": 1.8,
      "learning_rate": 2e-05,
      "loss": 0.3,
      "step": 591
    },
    {
      "epoch": 1.8,
      "learning_rate": 2e-05,
      "loss": 0.288,
      "step": 592
    },
    {
      "epoch": 1.8,
      "learning_rate": 2e-05,
      "loss": 0.2943,
      "step": 593
    },
    {
      "epoch": 1.81,
      "learning_rate": 2e-05,
      "loss": 0.2985,
      "step": 594
    },
    {
      "epoch": 1.81,
      "learning_rate": 2e-05,
      "loss": 0.3034,
      "step": 595
    },
    {
      "epoch": 1.81,
      "learning_rate": 2e-05,
      "loss": 0.2846,
      "step": 596
    },
    {
      "epoch": 1.81,
      "learning_rate": 2e-05,
      "loss": 0.2892,
      "step": 597
    },
    {
      "epoch": 1.82,
      "learning_rate": 2e-05,
      "loss": 0.286,
      "step": 598
    },
    {
      "epoch": 1.82,
      "learning_rate": 2e-05,
      "loss": 0.2851,
      "step": 599
    },
    {
      "epoch": 1.82,
      "learning_rate": 2e-05,
      "loss": 0.286,
      "step": 600
    },
    {
      "epoch": 1.83,
      "learning_rate": 2e-05,
      "loss": 0.3009,
      "step": 601
    },
    {
      "epoch": 1.83,
      "learning_rate": 2e-05,
      "loss": 0.3035,
      "step": 602
    },
    {
      "epoch": 1.83,
      "learning_rate": 2e-05,
      "loss": 0.3055,
      "step": 603
    },
    {
      "epoch": 1.84,
      "learning_rate": 2e-05,
      "loss": 0.2977,
      "step": 604
    },
    {
      "epoch": 1.84,
      "learning_rate": 2e-05,
      "loss": 0.2954,
      "step": 605
    },
    {
      "epoch": 1.84,
      "learning_rate": 2e-05,
      "loss": 0.2873,
      "step": 606
    },
    {
      "epoch": 1.84,
      "learning_rate": 2e-05,
      "loss": 0.2827,
      "step": 607
    },
    {
      "epoch": 1.85,
      "learning_rate": 2e-05,
      "loss": 0.2706,
      "step": 608
    },
    {
      "epoch": 1.85,
      "learning_rate": 2e-05,
      "loss": 0.2828,
      "step": 609
    },
    {
      "epoch": 1.85,
      "learning_rate": 2e-05,
      "loss": 0.2951,
      "step": 610
    },
    {
      "epoch": 1.86,
      "learning_rate": 2e-05,
      "loss": 0.2893,
      "step": 611
    },
    {
      "epoch": 1.86,
      "learning_rate": 2e-05,
      "loss": 0.2843,
      "step": 612
    },
    {
      "epoch": 1.86,
      "learning_rate": 2e-05,
      "loss": 0.2891,
      "step": 613
    },
    {
      "epoch": 1.87,
      "learning_rate": 2e-05,
      "loss": 0.2793,
      "step": 614
    },
    {
      "epoch": 1.87,
      "learning_rate": 2e-05,
      "loss": 0.28,
      "step": 615
    },
    {
      "epoch": 1.87,
      "learning_rate": 2e-05,
      "loss": 0.2902,
      "step": 616
    },
    {
      "epoch": 1.88,
      "learning_rate": 2e-05,
      "loss": 0.2924,
      "step": 617
    },
    {
      "epoch": 1.88,
      "learning_rate": 2e-05,
      "loss": 0.3005,
      "step": 618
    },
    {
      "epoch": 1.88,
      "learning_rate": 2e-05,
      "loss": 0.2745,
      "step": 619
    },
    {
      "epoch": 1.88,
      "learning_rate": 2e-05,
      "loss": 0.2987,
      "step": 620
    },
    {
      "epoch": 1.89,
      "learning_rate": 2e-05,
      "loss": 0.2932,
      "step": 621
    },
    {
      "epoch": 1.89,
      "learning_rate": 2e-05,
      "loss": 0.2788,
      "step": 622
    },
    {
      "epoch": 1.89,
      "learning_rate": 2e-05,
      "loss": 0.2795,
      "step": 623
    },
    {
      "epoch": 1.9,
      "learning_rate": 2e-05,
      "loss": 0.297,
      "step": 624
    },
    {
      "epoch": 1.9,
      "learning_rate": 2e-05,
      "loss": 0.2844,
      "step": 625
    },
    {
      "epoch": 1.9,
      "learning_rate": 2e-05,
      "loss": 0.2913,
      "step": 626
    },
    {
      "epoch": 1.91,
      "learning_rate": 2e-05,
      "loss": 0.2857,
      "step": 627
    },
    {
      "epoch": 1.91,
      "learning_rate": 2e-05,
      "loss": 0.2861,
      "step": 628
    },
    {
      "epoch": 1.91,
      "learning_rate": 2e-05,
      "loss": 0.277,
      "step": 629
    },
    {
      "epoch": 1.91,
      "learning_rate": 2e-05,
      "loss": 0.2874,
      "step": 630
    },
    {
      "epoch": 1.92,
      "learning_rate": 2e-05,
      "loss": 0.2808,
      "step": 631
    },
    {
      "epoch": 1.92,
      "learning_rate": 2e-05,
      "loss": 0.2914,
      "step": 632
    },
    {
      "epoch": 1.92,
      "learning_rate": 2e-05,
      "loss": 0.2848,
      "step": 633
    },
    {
      "epoch": 1.93,
      "learning_rate": 2e-05,
      "loss": 0.2872,
      "step": 634
    },
    {
      "epoch": 1.93,
      "learning_rate": 2e-05,
      "loss": 0.2737,
      "step": 635
    },
    {
      "epoch": 1.93,
      "learning_rate": 2e-05,
      "loss": 0.2752,
      "step": 636
    },
    {
      "epoch": 1.94,
      "learning_rate": 2e-05,
      "loss": 0.2853,
      "step": 637
    },
    {
      "epoch": 1.94,
      "learning_rate": 2e-05,
      "loss": 0.2846,
      "step": 638
    },
    {
      "epoch": 1.94,
      "learning_rate": 2e-05,
      "loss": 0.2821,
      "step": 639
    },
    {
      "epoch": 1.95,
      "learning_rate": 2e-05,
      "loss": 0.2927,
      "step": 640
    },
    {
      "epoch": 1.95,
      "learning_rate": 2e-05,
      "loss": 0.3013,
      "step": 641
    },
    {
      "epoch": 1.95,
      "learning_rate": 2e-05,
      "loss": 0.2881,
      "step": 642
    },
    {
      "epoch": 1.95,
      "learning_rate": 2e-05,
      "loss": 0.2789,
      "step": 643
    },
    {
      "epoch": 1.96,
      "learning_rate": 2e-05,
      "loss": 0.2882,
      "step": 644
    },
    {
      "epoch": 1.96,
      "learning_rate": 2e-05,
      "loss": 0.2769,
      "step": 645
    },
    {
      "epoch": 1.96,
      "learning_rate": 2e-05,
      "loss": 0.2722,
      "step": 646
    },
    {
      "epoch": 1.97,
      "learning_rate": 2e-05,
      "loss": 0.2899,
      "step": 647
    },
    {
      "epoch": 1.97,
      "learning_rate": 2e-05,
      "loss": 0.2831,
      "step": 648
    },
    {
      "epoch": 1.97,
      "learning_rate": 2e-05,
      "loss": 0.2925,
      "step": 649
    },
    {
      "epoch": 1.98,
      "learning_rate": 2e-05,
      "loss": 0.2769,
      "step": 650
    },
    {
      "epoch": 1.98,
      "learning_rate": 2e-05,
      "loss": 0.2744,
      "step": 651
    },
    {
      "epoch": 1.98,
      "learning_rate": 2e-05,
      "loss": 0.27,
      "step": 652
    },
    {
      "epoch": 1.98,
      "learning_rate": 2e-05,
      "loss": 0.2814,
      "step": 653
    },
    {
      "epoch": 1.99,
      "learning_rate": 2e-05,
      "loss": 0.2884,
      "step": 654
    },
    {
      "epoch": 1.99,
      "learning_rate": 2e-05,
      "loss": 0.2847,
      "step": 655
    },
    {
      "epoch": 1.99,
      "learning_rate": 2e-05,
      "loss": 0.2782,
      "step": 656
    },
    {
      "epoch": 2.0,
      "learning_rate": 2e-05,
      "loss": 0.2802,
      "step": 657
    },
    {
      "epoch": 2.0,
      "learning_rate": 2e-05,
      "loss": 0.2712,
      "step": 658
    },
    {
      "epoch": 2.0,
      "learning_rate": 2e-05,
      "loss": 0.2289,
      "step": 659
    },
    {
      "epoch": 2.01,
      "learning_rate": 2e-05,
      "loss": 0.2226,
      "step": 660
    },
    {
      "epoch": 2.01,
      "learning_rate": 2e-05,
      "loss": 0.2232,
      "step": 661
    },
    {
      "epoch": 2.01,
      "learning_rate": 2e-05,
      "loss": 0.2177,
      "step": 662
    },
    {
      "epoch": 2.02,
      "learning_rate": 2e-05,
      "loss": 0.2281,
      "step": 663
    },
    {
      "epoch": 2.02,
      "learning_rate": 2e-05,
      "loss": 0.2227,
      "step": 664
    },
    {
      "epoch": 2.02,
      "learning_rate": 2e-05,
      "loss": 0.2165,
      "step": 665
    },
    {
      "epoch": 2.02,
      "learning_rate": 2e-05,
      "loss": 0.2193,
      "step": 666
    },
    {
      "epoch": 2.03,
      "learning_rate": 2e-05,
      "loss": 0.2182,
      "step": 667
    },
    {
      "epoch": 2.03,
      "learning_rate": 2e-05,
      "loss": 0.2111,
      "step": 668
    },
    {
      "epoch": 2.03,
      "learning_rate": 2e-05,
      "loss": 0.2241,
      "step": 669
    },
    {
      "epoch": 2.04,
      "learning_rate": 2e-05,
      "loss": 0.2168,
      "step": 670
    },
    {
      "epoch": 2.04,
      "learning_rate": 2e-05,
      "loss": 0.2342,
      "step": 671
    },
    {
      "epoch": 2.04,
      "learning_rate": 2e-05,
      "loss": 0.2184,
      "step": 672
    },
    {
      "epoch": 2.05,
      "learning_rate": 2e-05,
      "loss": 0.2151,
      "step": 673
    },
    {
      "epoch": 2.05,
      "learning_rate": 2e-05,
      "loss": 0.2167,
      "step": 674
    },
    {
      "epoch": 2.05,
      "learning_rate": 2e-05,
      "loss": 0.2267,
      "step": 675
    },
    {
      "epoch": 2.05,
      "learning_rate": 2e-05,
      "loss": 0.2241,
      "step": 676
    },
    {
      "epoch": 2.06,
      "learning_rate": 2e-05,
      "loss": 0.2209,
      "step": 677
    },
    {
      "epoch": 2.06,
      "learning_rate": 2e-05,
      "loss": 0.2239,
      "step": 678
    },
    {
      "epoch": 2.06,
      "learning_rate": 2e-05,
      "loss": 0.2156,
      "step": 679
    },
    {
      "epoch": 2.07,
      "learning_rate": 2e-05,
      "loss": 0.2238,
      "step": 680
    },
    {
      "epoch": 2.07,
      "learning_rate": 2e-05,
      "loss": 0.2148,
      "step": 681
    },
    {
      "epoch": 2.07,
      "learning_rate": 2e-05,
      "loss": 0.2254,
      "step": 682
    },
    {
      "epoch": 2.08,
      "learning_rate": 2e-05,
      "loss": 0.2201,
      "step": 683
    },
    {
      "epoch": 2.08,
      "learning_rate": 2e-05,
      "loss": 0.2187,
      "step": 684
    },
    {
      "epoch": 2.08,
      "learning_rate": 2e-05,
      "loss": 0.2162,
      "step": 685
    },
    {
      "epoch": 2.09,
      "learning_rate": 2e-05,
      "loss": 0.2194,
      "step": 686
    },
    {
      "epoch": 2.09,
      "learning_rate": 2e-05,
      "loss": 0.2183,
      "step": 687
    },
    {
      "epoch": 2.09,
      "learning_rate": 2e-05,
      "loss": 0.2227,
      "step": 688
    },
    {
      "epoch": 2.09,
      "learning_rate": 2e-05,
      "loss": 0.2154,
      "step": 689
    },
    {
      "epoch": 2.1,
      "learning_rate": 2e-05,
      "loss": 0.2196,
      "step": 690
    },
    {
      "epoch": 2.1,
      "learning_rate": 2e-05,
      "loss": 0.2274,
      "step": 691
    },
    {
      "epoch": 2.1,
      "learning_rate": 2e-05,
      "loss": 0.2078,
      "step": 692
    },
    {
      "epoch": 2.11,
      "learning_rate": 2e-05,
      "loss": 0.2161,
      "step": 693
    },
    {
      "epoch": 2.11,
      "learning_rate": 2e-05,
      "loss": 0.2285,
      "step": 694
    },
    {
      "epoch": 2.11,
      "learning_rate": 2e-05,
      "loss": 0.2259,
      "step": 695
    },
    {
      "epoch": 2.12,
      "learning_rate": 2e-05,
      "loss": 0.2257,
      "step": 696
    },
    {
      "epoch": 2.12,
      "learning_rate": 2e-05,
      "loss": 0.2283,
      "step": 697
    },
    {
      "epoch": 2.12,
      "learning_rate": 2e-05,
      "loss": 0.2188,
      "step": 698
    },
    {
      "epoch": 2.12,
      "learning_rate": 2e-05,
      "loss": 0.2214,
      "step": 699
    },
    {
      "epoch": 2.13,
      "learning_rate": 2e-05,
      "loss": 0.2161,
      "step": 700
    },
    {
      "epoch": 2.13,
      "learning_rate": 2e-05,
      "loss": 0.2314,
      "step": 701
    },
    {
      "epoch": 2.13,
      "learning_rate": 2e-05,
      "loss": 0.2299,
      "step": 702
    },
    {
      "epoch": 2.14,
      "learning_rate": 2e-05,
      "loss": 0.223,
      "step": 703
    },
    {
      "epoch": 2.14,
      "learning_rate": 2e-05,
      "loss": 0.2258,
      "step": 704
    },
    {
      "epoch": 2.14,
      "learning_rate": 2e-05,
      "loss": 0.2157,
      "step": 705
    },
    {
      "epoch": 2.15,
      "learning_rate": 2e-05,
      "loss": 0.2091,
      "step": 706
    },
    {
      "epoch": 2.15,
      "learning_rate": 2e-05,
      "loss": 0.2214,
      "step": 707
    },
    {
      "epoch": 2.15,
      "learning_rate": 2e-05,
      "loss": 0.2235,
      "step": 708
    },
    {
      "epoch": 2.16,
      "learning_rate": 2e-05,
      "loss": 0.2276,
      "step": 709
    },
    {
      "epoch": 2.16,
      "learning_rate": 2e-05,
      "loss": 0.2142,
      "step": 710
    },
    {
      "epoch": 2.16,
      "learning_rate": 2e-05,
      "loss": 0.2284,
      "step": 711
    },
    {
      "epoch": 2.16,
      "learning_rate": 2e-05,
      "loss": 0.2114,
      "step": 712
    },
    {
      "epoch": 2.17,
      "learning_rate": 2e-05,
      "loss": 0.2182,
      "step": 713
    },
    {
      "epoch": 2.17,
      "learning_rate": 2e-05,
      "loss": 0.2249,
      "step": 714
    },
    {
      "epoch": 2.17,
      "learning_rate": 2e-05,
      "loss": 0.2178,
      "step": 715
    },
    {
      "epoch": 2.18,
      "learning_rate": 2e-05,
      "loss": 0.2321,
      "step": 716
    },
    {
      "epoch": 2.18,
      "learning_rate": 2e-05,
      "loss": 0.2255,
      "step": 717
    },
    {
      "epoch": 2.18,
      "learning_rate": 2e-05,
      "loss": 0.2137,
      "step": 718
    },
    {
      "epoch": 2.19,
      "learning_rate": 2e-05,
      "loss": 0.2245,
      "step": 719
    },
    {
      "epoch": 2.19,
      "learning_rate": 2e-05,
      "loss": 0.2206,
      "step": 720
    },
    {
      "epoch": 2.19,
      "learning_rate": 2e-05,
      "loss": 0.2175,
      "step": 721
    },
    {
      "epoch": 2.19,
      "learning_rate": 2e-05,
      "loss": 0.225,
      "step": 722
    },
    {
      "epoch": 2.2,
      "learning_rate": 2e-05,
      "loss": 0.2283,
      "step": 723
    },
    {
      "epoch": 2.2,
      "learning_rate": 2e-05,
      "loss": 0.2189,
      "step": 724
    },
    {
      "epoch": 2.2,
      "learning_rate": 2e-05,
      "loss": 0.2251,
      "step": 725
    },
    {
      "epoch": 2.21,
      "learning_rate": 2e-05,
      "loss": 0.2206,
      "step": 726
    },
    {
      "epoch": 2.21,
      "learning_rate": 2e-05,
      "loss": 0.2138,
      "step": 727
    },
    {
      "epoch": 2.21,
      "learning_rate": 2e-05,
      "loss": 0.2155,
      "step": 728
    },
    {
      "epoch": 2.22,
      "learning_rate": 2e-05,
      "loss": 0.2243,
      "step": 729
    },
    {
      "epoch": 2.22,
      "learning_rate": 2e-05,
      "loss": 0.2234,
      "step": 730
    },
    {
      "epoch": 2.22,
      "learning_rate": 2e-05,
      "loss": 0.206,
      "step": 731
    },
    {
      "epoch": 2.22,
      "learning_rate": 2e-05,
      "loss": 0.2285,
      "step": 732
    },
    {
      "epoch": 2.23,
      "learning_rate": 2e-05,
      "loss": 0.2208,
      "step": 733
    },
    {
      "epoch": 2.23,
      "learning_rate": 2e-05,
      "loss": 0.2233,
      "step": 734
    },
    {
      "epoch": 2.23,
      "learning_rate": 2e-05,
      "loss": 0.227,
      "step": 735
    },
    {
      "epoch": 2.24,
      "learning_rate": 2e-05,
      "loss": 0.2216,
      "step": 736
    },
    {
      "epoch": 2.24,
      "learning_rate": 2e-05,
      "loss": 0.2308,
      "step": 737
    },
    {
      "epoch": 2.24,
      "learning_rate": 2e-05,
      "loss": 0.2164,
      "step": 738
    },
    {
      "epoch": 2.25,
      "learning_rate": 2e-05,
      "loss": 0.2246,
      "step": 739
    },
    {
      "epoch": 2.25,
      "learning_rate": 2e-05,
      "loss": 0.2315,
      "step": 740
    },
    {
      "epoch": 2.25,
      "learning_rate": 2e-05,
      "loss": 0.2206,
      "step": 741
    },
    {
      "epoch": 2.26,
      "learning_rate": 2e-05,
      "loss": 0.2238,
      "step": 742
    },
    {
      "epoch": 2.26,
      "learning_rate": 2e-05,
      "loss": 0.2171,
      "step": 743
    },
    {
      "epoch": 2.26,
      "learning_rate": 2e-05,
      "loss": 0.2208,
      "step": 744
    },
    {
      "epoch": 2.26,
      "learning_rate": 2e-05,
      "loss": 0.2197,
      "step": 745
    },
    {
      "epoch": 2.27,
      "learning_rate": 2e-05,
      "loss": 0.2348,
      "step": 746
    },
    {
      "epoch": 2.27,
      "learning_rate": 2e-05,
      "loss": 0.2202,
      "step": 747
    },
    {
      "epoch": 2.27,
      "learning_rate": 2e-05,
      "loss": 0.215,
      "step": 748
    },
    {
      "epoch": 2.28,
      "learning_rate": 2e-05,
      "loss": 0.2113,
      "step": 749
    },
    {
      "epoch": 2.28,
      "learning_rate": 2e-05,
      "loss": 0.2122,
      "step": 750
    },
    {
      "epoch": 2.28,
      "learning_rate": 2e-05,
      "loss": 0.2349,
      "step": 751
    },
    {
      "epoch": 2.29,
      "learning_rate": 2e-05,
      "loss": 0.2206,
      "step": 752
    },
    {
      "epoch": 2.29,
      "learning_rate": 2e-05,
      "loss": 0.2139,
      "step": 753
    },
    {
      "epoch": 2.29,
      "learning_rate": 2e-05,
      "loss": 0.2177,
      "step": 754
    },
    {
      "epoch": 2.29,
      "learning_rate": 2e-05,
      "loss": 0.2173,
      "step": 755
    },
    {
      "epoch": 2.3,
      "learning_rate": 2e-05,
      "loss": 0.2139,
      "step": 756
    },
    {
      "epoch": 2.3,
      "learning_rate": 2e-05,
      "loss": 0.2143,
      "step": 757
    },
    {
      "epoch": 2.3,
      "learning_rate": 2e-05,
      "loss": 0.2283,
      "step": 758
    },
    {
      "epoch": 2.31,
      "learning_rate": 2e-05,
      "loss": 0.2115,
      "step": 759
    },
    {
      "epoch": 2.31,
      "learning_rate": 2e-05,
      "loss": 0.214,
      "step": 760
    },
    {
      "epoch": 2.31,
      "learning_rate": 2e-05,
      "loss": 0.2226,
      "step": 761
    },
    {
      "epoch": 2.32,
      "learning_rate": 2e-05,
      "loss": 0.2154,
      "step": 762
    },
    {
      "epoch": 2.32,
      "learning_rate": 2e-05,
      "loss": 0.2202,
      "step": 763
    },
    {
      "epoch": 2.32,
      "learning_rate": 2e-05,
      "loss": 0.2235,
      "step": 764
    },
    {
      "epoch": 2.33,
      "learning_rate": 2e-05,
      "loss": 0.2226,
      "step": 765
    },
    {
      "epoch": 2.33,
      "learning_rate": 2e-05,
      "loss": 0.2191,
      "step": 766
    },
    {
      "epoch": 2.33,
      "learning_rate": 2e-05,
      "loss": 0.216,
      "step": 767
    },
    {
      "epoch": 2.33,
      "learning_rate": 2e-05,
      "loss": 0.2192,
      "step": 768
    },
    {
      "epoch": 2.34,
      "learning_rate": 2e-05,
      "loss": 0.2309,
      "step": 769
    },
    {
      "epoch": 2.34,
      "learning_rate": 2e-05,
      "loss": 0.2225,
      "step": 770
    },
    {
      "epoch": 2.34,
      "learning_rate": 2e-05,
      "loss": 0.2276,
      "step": 771
    },
    {
      "epoch": 2.35,
      "learning_rate": 2e-05,
      "loss": 0.2196,
      "step": 772
    },
    {
      "epoch": 2.35,
      "learning_rate": 2e-05,
      "loss": 0.2167,
      "step": 773
    },
    {
      "epoch": 2.35,
      "learning_rate": 2e-05,
      "loss": 0.2235,
      "step": 774
    },
    {
      "epoch": 2.36,
      "learning_rate": 2e-05,
      "loss": 0.2209,
      "step": 775
    },
    {
      "epoch": 2.36,
      "learning_rate": 2e-05,
      "loss": 0.218,
      "step": 776
    },
    {
      "epoch": 2.36,
      "learning_rate": 2e-05,
      "loss": 0.2069,
      "step": 777
    },
    {
      "epoch": 2.36,
      "learning_rate": 2e-05,
      "loss": 0.2205,
      "step": 778
    },
    {
      "epoch": 2.37,
      "learning_rate": 2e-05,
      "loss": 0.2284,
      "step": 779
    },
    {
      "epoch": 2.37,
      "learning_rate": 2e-05,
      "loss": 0.2174,
      "step": 780
    },
    {
      "epoch": 2.37,
      "learning_rate": 2e-05,
      "loss": 0.2215,
      "step": 781
    },
    {
      "epoch": 2.38,
      "learning_rate": 2e-05,
      "loss": 0.2135,
      "step": 782
    },
    {
      "epoch": 2.38,
      "learning_rate": 2e-05,
      "loss": 0.2161,
      "step": 783
    },
    {
      "epoch": 2.38,
      "learning_rate": 2e-05,
      "loss": 0.2279,
      "step": 784
    },
    {
      "epoch": 2.39,
      "learning_rate": 2e-05,
      "loss": 0.2348,
      "step": 785
    },
    {
      "epoch": 2.39,
      "learning_rate": 2e-05,
      "loss": 0.2272,
      "step": 786
    },
    {
      "epoch": 2.39,
      "learning_rate": 2e-05,
      "loss": 0.2182,
      "step": 787
    },
    {
      "epoch": 2.4,
      "learning_rate": 2e-05,
      "loss": 0.2186,
      "step": 788
    },
    {
      "epoch": 2.4,
      "learning_rate": 2e-05,
      "loss": 0.2238,
      "step": 789
    },
    {
      "epoch": 2.4,
      "learning_rate": 2e-05,
      "loss": 0.2311,
      "step": 790
    },
    {
      "epoch": 2.4,
      "learning_rate": 2e-05,
      "loss": 0.2156,
      "step": 791
    },
    {
      "epoch": 2.41,
      "learning_rate": 2e-05,
      "loss": 0.2185,
      "step": 792
    },
    {
      "epoch": 2.41,
      "learning_rate": 2e-05,
      "loss": 0.217,
      "step": 793
    },
    {
      "epoch": 2.41,
      "learning_rate": 2e-05,
      "loss": 0.2153,
      "step": 794
    },
    {
      "epoch": 2.42,
      "learning_rate": 2e-05,
      "loss": 0.2161,
      "step": 795
    },
    {
      "epoch": 2.42,
      "learning_rate": 2e-05,
      "loss": 0.2187,
      "step": 796
    },
    {
      "epoch": 2.42,
      "learning_rate": 2e-05,
      "loss": 0.2147,
      "step": 797
    },
    {
      "epoch": 2.43,
      "learning_rate": 2e-05,
      "loss": 0.2187,
      "step": 798
    },
    {
      "epoch": 2.43,
      "learning_rate": 2e-05,
      "loss": 0.2287,
      "step": 799
    },
    {
      "epoch": 2.43,
      "learning_rate": 2e-05,
      "loss": 0.2293,
      "step": 800
    },
    {
      "epoch": 2.43,
      "learning_rate": 2e-05,
      "loss": 0.2204,
      "step": 801
    },
    {
      "epoch": 2.44,
      "learning_rate": 2e-05,
      "loss": 0.2251,
      "step": 802
    },
    {
      "epoch": 2.44,
      "learning_rate": 2e-05,
      "loss": 0.224,
      "step": 803
    },
    {
      "epoch": 2.44,
      "learning_rate": 2e-05,
      "loss": 0.2236,
      "step": 804
    },
    {
      "epoch": 2.45,
      "learning_rate": 2e-05,
      "loss": 0.2219,
      "step": 805
    },
    {
      "epoch": 2.45,
      "learning_rate": 2e-05,
      "loss": 0.2219,
      "step": 806
    },
    {
      "epoch": 2.45,
      "learning_rate": 2e-05,
      "loss": 0.2174,
      "step": 807
    },
    {
      "epoch": 2.46,
      "learning_rate": 2e-05,
      "loss": 0.218,
      "step": 808
    },
    {
      "epoch": 2.46,
      "learning_rate": 2e-05,
      "loss": 0.214,
      "step": 809
    },
    {
      "epoch": 2.46,
      "learning_rate": 2e-05,
      "loss": 0.2175,
      "step": 810
    },
    {
      "epoch": 2.47,
      "learning_rate": 2e-05,
      "loss": 0.2121,
      "step": 811
    },
    {
      "epoch": 2.47,
      "learning_rate": 2e-05,
      "loss": 0.2158,
      "step": 812
    },
    {
      "epoch": 2.47,
      "learning_rate": 2e-05,
      "loss": 0.2182,
      "step": 813
    },
    {
      "epoch": 2.47,
      "learning_rate": 2e-05,
      "loss": 0.2163,
      "step": 814
    },
    {
      "epoch": 2.48,
      "learning_rate": 2e-05,
      "loss": 0.2114,
      "step": 815
    },
    {
      "epoch": 2.48,
      "learning_rate": 2e-05,
      "loss": 0.2251,
      "step": 816
    },
    {
      "epoch": 2.48,
      "learning_rate": 2e-05,
      "loss": 0.2165,
      "step": 817
    },
    {
      "epoch": 2.49,
      "learning_rate": 2e-05,
      "loss": 0.2093,
      "step": 818
    },
    {
      "epoch": 2.49,
      "learning_rate": 2e-05,
      "loss": 0.2163,
      "step": 819
    },
    {
      "epoch": 2.49,
      "learning_rate": 2e-05,
      "loss": 0.2203,
      "step": 820
    },
    {
      "epoch": 2.5,
      "learning_rate": 2e-05,
      "loss": 0.2244,
      "step": 821
    },
    {
      "epoch": 2.5,
      "learning_rate": 2e-05,
      "loss": 0.2227,
      "step": 822
    },
    {
      "epoch": 2.5,
      "learning_rate": 2e-05,
      "loss": 0.2139,
      "step": 823
    },
    {
      "epoch": 2.5,
      "learning_rate": 2e-05,
      "loss": 0.2146,
      "step": 824
    },
    {
      "epoch": 2.51,
      "learning_rate": 2e-05,
      "loss": 0.2141,
      "step": 825
    },
    {
      "epoch": 2.51,
      "learning_rate": 2e-05,
      "loss": 0.2245,
      "step": 826
    },
    {
      "epoch": 2.51,
      "learning_rate": 2e-05,
      "loss": 0.2271,
      "step": 827
    },
    {
      "epoch": 2.52,
      "learning_rate": 2e-05,
      "loss": 0.2206,
      "step": 828
    },
    {
      "epoch": 2.52,
      "learning_rate": 2e-05,
      "loss": 0.2159,
      "step": 829
    },
    {
      "epoch": 2.52,
      "learning_rate": 2e-05,
      "loss": 0.2208,
      "step": 830
    },
    {
      "epoch": 2.53,
      "learning_rate": 2e-05,
      "loss": 0.2161,
      "step": 831
    },
    {
      "epoch": 2.53,
      "learning_rate": 2e-05,
      "loss": 0.2208,
      "step": 832
    },
    {
      "epoch": 2.53,
      "learning_rate": 2e-05,
      "loss": 0.2154,
      "step": 833
    },
    {
      "epoch": 2.53,
      "learning_rate": 2e-05,
      "loss": 0.2262,
      "step": 834
    },
    {
      "epoch": 2.54,
      "learning_rate": 2e-05,
      "loss": 0.2244,
      "step": 835
    },
    {
      "epoch": 2.54,
      "learning_rate": 2e-05,
      "loss": 0.2172,
      "step": 836
    },
    {
      "epoch": 2.54,
      "learning_rate": 2e-05,
      "loss": 0.222,
      "step": 837
    },
    {
      "epoch": 2.55,
      "learning_rate": 2e-05,
      "loss": 0.2259,
      "step": 838
    },
    {
      "epoch": 2.55,
      "learning_rate": 2e-05,
      "loss": 0.2212,
      "step": 839
    },
    {
      "epoch": 2.55,
      "learning_rate": 2e-05,
      "loss": 0.2193,
      "step": 840
    },
    {
      "epoch": 2.56,
      "learning_rate": 2e-05,
      "loss": 0.2236,
      "step": 841
    },
    {
      "epoch": 2.56,
      "learning_rate": 2e-05,
      "loss": 0.2179,
      "step": 842
    },
    {
      "epoch": 2.56,
      "learning_rate": 2e-05,
      "loss": 0.2237,
      "step": 843
    },
    {
      "epoch": 2.57,
      "learning_rate": 2e-05,
      "loss": 0.2081,
      "step": 844
    },
    {
      "epoch": 2.57,
      "learning_rate": 2e-05,
      "loss": 0.2148,
      "step": 845
    },
    {
      "epoch": 2.57,
      "learning_rate": 2e-05,
      "loss": 0.2158,
      "step": 846
    },
    {
      "epoch": 2.57,
      "learning_rate": 2e-05,
      "loss": 0.213,
      "step": 847
    },
    {
      "epoch": 2.58,
      "learning_rate": 2e-05,
      "loss": 0.2174,
      "step": 848
    },
    {
      "epoch": 2.58,
      "learning_rate": 2e-05,
      "loss": 0.2157,
      "step": 849
    },
    {
      "epoch": 2.58,
      "learning_rate": 2e-05,
      "loss": 0.2069,
      "step": 850
    },
    {
      "epoch": 2.59,
      "learning_rate": 2e-05,
      "loss": 0.2108,
      "step": 851
    },
    {
      "epoch": 2.59,
      "learning_rate": 2e-05,
      "loss": 0.212,
      "step": 852
    },
    {
      "epoch": 2.59,
      "learning_rate": 2e-05,
      "loss": 0.2137,
      "step": 853
    },
    {
      "epoch": 2.6,
      "learning_rate": 2e-05,
      "loss": 0.2192,
      "step": 854
    },
    {
      "epoch": 2.6,
      "learning_rate": 2e-05,
      "loss": 0.2201,
      "step": 855
    },
    {
      "epoch": 2.6,
      "learning_rate": 2e-05,
      "loss": 0.2253,
      "step": 856
    },
    {
      "epoch": 2.6,
      "learning_rate": 2e-05,
      "loss": 0.2134,
      "step": 857
    },
    {
      "epoch": 2.61,
      "learning_rate": 2e-05,
      "loss": 0.2114,
      "step": 858
    },
    {
      "epoch": 2.61,
      "learning_rate": 2e-05,
      "loss": 0.2188,
      "step": 859
    },
    {
      "epoch": 2.61,
      "learning_rate": 2e-05,
      "loss": 0.2196,
      "step": 860
    },
    {
      "epoch": 2.62,
      "learning_rate": 2e-05,
      "loss": 0.2157,
      "step": 861
    },
    {
      "epoch": 2.62,
      "learning_rate": 2e-05,
      "loss": 0.2144,
      "step": 862
    },
    {
      "epoch": 2.62,
      "learning_rate": 2e-05,
      "loss": 0.2194,
      "step": 863
    },
    {
      "epoch": 2.63,
      "learning_rate": 2e-05,
      "loss": 0.2162,
      "step": 864
    },
    {
      "epoch": 2.63,
      "learning_rate": 2e-05,
      "loss": 0.212,
      "step": 865
    },
    {
      "epoch": 2.63,
      "learning_rate": 2e-05,
      "loss": 0.2204,
      "step": 866
    },
    {
      "epoch": 2.64,
      "learning_rate": 2e-05,
      "loss": 0.2208,
      "step": 867
    },
    {
      "epoch": 2.64,
      "learning_rate": 2e-05,
      "loss": 0.2165,
      "step": 868
    },
    {
      "epoch": 2.64,
      "learning_rate": 2e-05,
      "loss": 0.2277,
      "step": 869
    },
    {
      "epoch": 2.64,
      "learning_rate": 2e-05,
      "loss": 0.2067,
      "step": 870
    },
    {
      "epoch": 2.65,
      "learning_rate": 2e-05,
      "loss": 0.2095,
      "step": 871
    },
    {
      "epoch": 2.65,
      "learning_rate": 2e-05,
      "loss": 0.2121,
      "step": 872
    },
    {
      "epoch": 2.65,
      "learning_rate": 2e-05,
      "loss": 0.2318,
      "step": 873
    },
    {
      "epoch": 2.66,
      "learning_rate": 2e-05,
      "loss": 0.2131,
      "step": 874
    },
    {
      "epoch": 2.66,
      "learning_rate": 2e-05,
      "loss": 0.2192,
      "step": 875
    },
    {
      "epoch": 2.66,
      "learning_rate": 2e-05,
      "loss": 0.2143,
      "step": 876
    },
    {
      "epoch": 2.67,
      "learning_rate": 2e-05,
      "loss": 0.221,
      "step": 877
    },
    {
      "epoch": 2.67,
      "learning_rate": 2e-05,
      "loss": 0.2205,
      "step": 878
    },
    {
      "epoch": 2.67,
      "learning_rate": 2e-05,
      "loss": 0.2072,
      "step": 879
    },
    {
      "epoch": 2.67,
      "learning_rate": 2e-05,
      "loss": 0.2129,
      "step": 880
    },
    {
      "epoch": 2.68,
      "learning_rate": 2e-05,
      "loss": 0.2127,
      "step": 881
    },
    {
      "epoch": 2.68,
      "learning_rate": 2e-05,
      "loss": 0.2111,
      "step": 882
    },
    {
      "epoch": 2.68,
      "learning_rate": 2e-05,
      "loss": 0.21,
      "step": 883
    },
    {
      "epoch": 2.69,
      "learning_rate": 2e-05,
      "loss": 0.2193,
      "step": 884
    },
    {
      "epoch": 2.69,
      "learning_rate": 2e-05,
      "loss": 0.2115,
      "step": 885
    },
    {
      "epoch": 2.69,
      "learning_rate": 2e-05,
      "loss": 0.2102,
      "step": 886
    },
    {
      "epoch": 2.7,
      "learning_rate": 2e-05,
      "loss": 0.2146,
      "step": 887
    },
    {
      "epoch": 2.7,
      "learning_rate": 2e-05,
      "loss": 0.2036,
      "step": 888
    },
    {
      "epoch": 2.7,
      "learning_rate": 2e-05,
      "loss": 0.2196,
      "step": 889
    },
    {
      "epoch": 2.71,
      "learning_rate": 2e-05,
      "loss": 0.208,
      "step": 890
    },
    {
      "epoch": 2.71,
      "learning_rate": 2e-05,
      "loss": 0.2159,
      "step": 891
    },
    {
      "epoch": 2.71,
      "learning_rate": 2e-05,
      "loss": 0.2228,
      "step": 892
    },
    {
      "epoch": 2.71,
      "learning_rate": 2e-05,
      "loss": 0.2185,
      "step": 893
    },
    {
      "epoch": 2.72,
      "learning_rate": 2e-05,
      "loss": 0.2088,
      "step": 894
    },
    {
      "epoch": 2.72,
      "learning_rate": 2e-05,
      "loss": 0.2108,
      "step": 895
    },
    {
      "epoch": 2.72,
      "learning_rate": 2e-05,
      "loss": 0.2162,
      "step": 896
    },
    {
      "epoch": 2.73,
      "learning_rate": 2e-05,
      "loss": 0.2143,
      "step": 897
    },
    {
      "epoch": 2.73,
      "learning_rate": 2e-05,
      "loss": 0.2167,
      "step": 898
    },
    {
      "epoch": 2.73,
      "learning_rate": 2e-05,
      "loss": 0.2066,
      "step": 899
    },
    {
      "epoch": 2.74,
      "learning_rate": 2e-05,
      "loss": 0.2104,
      "step": 900
    },
    {
      "epoch": 2.74,
      "learning_rate": 2e-05,
      "loss": 0.2202,
      "step": 901
    },
    {
      "epoch": 2.74,
      "learning_rate": 2e-05,
      "loss": 0.2156,
      "step": 902
    },
    {
      "epoch": 2.74,
      "learning_rate": 2e-05,
      "loss": 0.2168,
      "step": 903
    },
    {
      "epoch": 2.75,
      "learning_rate": 2e-05,
      "loss": 0.2132,
      "step": 904
    },
    {
      "epoch": 2.75,
      "learning_rate": 2e-05,
      "loss": 0.2111,
      "step": 905
    },
    {
      "epoch": 2.75,
      "learning_rate": 2e-05,
      "loss": 0.2168,
      "step": 906
    },
    {
      "epoch": 2.76,
      "learning_rate": 2e-05,
      "loss": 0.211,
      "step": 907
    },
    {
      "epoch": 2.76,
      "learning_rate": 2e-05,
      "loss": 0.2153,
      "step": 908
    },
    {
      "epoch": 2.76,
      "learning_rate": 2e-05,
      "loss": 0.209,
      "step": 909
    },
    {
      "epoch": 2.77,
      "learning_rate": 2e-05,
      "loss": 0.2135,
      "step": 910
    },
    {
      "epoch": 2.77,
      "learning_rate": 2e-05,
      "loss": 0.211,
      "step": 911
    },
    {
      "epoch": 2.77,
      "learning_rate": 2e-05,
      "loss": 0.2175,
      "step": 912
    },
    {
      "epoch": 2.78,
      "learning_rate": 2e-05,
      "loss": 0.2069,
      "step": 913
    },
    {
      "epoch": 2.78,
      "learning_rate": 2e-05,
      "loss": 0.2176,
      "step": 914
    },
    {
      "epoch": 2.78,
      "learning_rate": 2e-05,
      "loss": 0.2081,
      "step": 915
    },
    {
      "epoch": 2.78,
      "learning_rate": 2e-05,
      "loss": 0.2077,
      "step": 916
    },
    {
      "epoch": 2.79,
      "learning_rate": 2e-05,
      "loss": 0.2114,
      "step": 917
    },
    {
      "epoch": 2.79,
      "learning_rate": 2e-05,
      "loss": 0.2169,
      "step": 918
    },
    {
      "epoch": 2.79,
      "learning_rate": 2e-05,
      "loss": 0.2107,
      "step": 919
    },
    {
      "epoch": 2.8,
      "learning_rate": 2e-05,
      "loss": 0.2178,
      "step": 920
    },
    {
      "epoch": 2.8,
      "learning_rate": 2e-05,
      "loss": 0.214,
      "step": 921
    },
    {
      "epoch": 2.8,
      "learning_rate": 2e-05,
      "loss": 0.2129,
      "step": 922
    },
    {
      "epoch": 2.81,
      "learning_rate": 2e-05,
      "loss": 0.2243,
      "step": 923
    },
    {
      "epoch": 2.81,
      "learning_rate": 2e-05,
      "loss": 0.215,
      "step": 924
    },
    {
      "epoch": 2.81,
      "learning_rate": 2e-05,
      "loss": 0.2085,
      "step": 925
    },
    {
      "epoch": 2.81,
      "learning_rate": 2e-05,
      "loss": 0.2138,
      "step": 926
    },
    {
      "epoch": 2.82,
      "learning_rate": 2e-05,
      "loss": 0.1992,
      "step": 927
    },
    {
      "epoch": 2.82,
      "learning_rate": 2e-05,
      "loss": 0.2163,
      "step": 928
    },
    {
      "epoch": 2.82,
      "learning_rate": 2e-05,
      "loss": 0.2041,
      "step": 929
    },
    {
      "epoch": 2.83,
      "learning_rate": 2e-05,
      "loss": 0.2188,
      "step": 930
    },
    {
      "epoch": 2.83,
      "learning_rate": 2e-05,
      "loss": 0.2133,
      "step": 931
    },
    {
      "epoch": 2.83,
      "learning_rate": 2e-05,
      "loss": 0.2154,
      "step": 932
    },
    {
      "epoch": 2.84,
      "learning_rate": 2e-05,
      "loss": 0.2184,
      "step": 933
    },
    {
      "epoch": 2.84,
      "learning_rate": 2e-05,
      "loss": 0.2034,
      "step": 934
    },
    {
      "epoch": 2.84,
      "learning_rate": 2e-05,
      "loss": 0.2128,
      "step": 935
    },
    {
      "epoch": 2.84,
      "learning_rate": 2e-05,
      "loss": 0.208,
      "step": 936
    },
    {
      "epoch": 2.85,
      "learning_rate": 2e-05,
      "loss": 0.2165,
      "step": 937
    },
    {
      "epoch": 2.85,
      "learning_rate": 2e-05,
      "loss": 0.1996,
      "step": 938
    },
    {
      "epoch": 2.85,
      "learning_rate": 2e-05,
      "loss": 0.2185,
      "step": 939
    },
    {
      "epoch": 2.86,
      "learning_rate": 2e-05,
      "loss": 0.2097,
      "step": 940
    },
    {
      "epoch": 2.86,
      "learning_rate": 2e-05,
      "loss": 0.2025,
      "step": 941
    },
    {
      "epoch": 2.86,
      "learning_rate": 2e-05,
      "loss": 0.1997,
      "step": 942
    },
    {
      "epoch": 2.87,
      "learning_rate": 2e-05,
      "loss": 0.2122,
      "step": 943
    },
    {
      "epoch": 2.87,
      "learning_rate": 2e-05,
      "loss": 0.2061,
      "step": 944
    },
    {
      "epoch": 2.87,
      "learning_rate": 2e-05,
      "loss": 0.2153,
      "step": 945
    },
    {
      "epoch": 2.88,
      "learning_rate": 2e-05,
      "loss": 0.2078,
      "step": 946
    },
    {
      "epoch": 2.88,
      "learning_rate": 2e-05,
      "loss": 0.2071,
      "step": 947
    },
    {
      "epoch": 2.88,
      "learning_rate": 2e-05,
      "loss": 0.2123,
      "step": 948
    },
    {
      "epoch": 2.88,
      "learning_rate": 2e-05,
      "loss": 0.2141,
      "step": 949
    },
    {
      "epoch": 2.89,
      "learning_rate": 2e-05,
      "loss": 0.2169,
      "step": 950
    },
    {
      "epoch": 2.89,
      "learning_rate": 2e-05,
      "loss": 0.2039,
      "step": 951
    },
    {
      "epoch": 2.89,
      "learning_rate": 2e-05,
      "loss": 0.2062,
      "step": 952
    },
    {
      "epoch": 2.9,
      "learning_rate": 2e-05,
      "loss": 0.2106,
      "step": 953
    },
    {
      "epoch": 2.9,
      "learning_rate": 2e-05,
      "loss": 0.1998,
      "step": 954
    },
    {
      "epoch": 2.9,
      "learning_rate": 2e-05,
      "loss": 0.2224,
      "step": 955
    },
    {
      "epoch": 2.91,
      "learning_rate": 2e-05,
      "loss": 0.2075,
      "step": 956
    },
    {
      "epoch": 2.91,
      "learning_rate": 2e-05,
      "loss": 0.2215,
      "step": 957
    },
    {
      "epoch": 2.91,
      "learning_rate": 2e-05,
      "loss": 0.2037,
      "step": 958
    },
    {
      "epoch": 2.91,
      "learning_rate": 2e-05,
      "loss": 0.2041,
      "step": 959
    },
    {
      "epoch": 2.92,
      "learning_rate": 2e-05,
      "loss": 0.2093,
      "step": 960
    },
    {
      "epoch": 2.92,
      "learning_rate": 2e-05,
      "loss": 0.2088,
      "step": 961
    },
    {
      "epoch": 2.92,
      "learning_rate": 2e-05,
      "loss": 0.2007,
      "step": 962
    },
    {
      "epoch": 2.93,
      "learning_rate": 2e-05,
      "loss": 0.2075,
      "step": 963
    },
    {
      "epoch": 2.93,
      "learning_rate": 2e-05,
      "loss": 0.2101,
      "step": 964
    },
    {
      "epoch": 2.93,
      "learning_rate": 2e-05,
      "loss": 0.2121,
      "step": 965
    },
    {
      "epoch": 2.94,
      "learning_rate": 2e-05,
      "loss": 0.2131,
      "step": 966
    },
    {
      "epoch": 2.94,
      "learning_rate": 2e-05,
      "loss": 0.2124,
      "step": 967
    },
    {
      "epoch": 2.94,
      "learning_rate": 2e-05,
      "loss": 0.209,
      "step": 968
    },
    {
      "epoch": 2.95,
      "learning_rate": 2e-05,
      "loss": 0.2039,
      "step": 969
    },
    {
      "epoch": 2.95,
      "learning_rate": 2e-05,
      "loss": 0.2075,
      "step": 970
    },
    {
      "epoch": 2.95,
      "learning_rate": 2e-05,
      "loss": 0.2125,
      "step": 971
    },
    {
      "epoch": 2.95,
      "learning_rate": 2e-05,
      "loss": 0.2092,
      "step": 972
    },
    {
      "epoch": 2.96,
      "learning_rate": 2e-05,
      "loss": 0.2115,
      "step": 973
    },
    {
      "epoch": 2.96,
      "learning_rate": 2e-05,
      "loss": 0.2108,
      "step": 974
    },
    {
      "epoch": 2.96,
      "learning_rate": 2e-05,
      "loss": 0.2016,
      "step": 975
    },
    {
      "epoch": 2.97,
      "learning_rate": 2e-05,
      "loss": 0.2065,
      "step": 976
    },
    {
      "epoch": 2.97,
      "learning_rate": 2e-05,
      "loss": 0.2065,
      "step": 977
    },
    {
      "epoch": 2.97,
      "learning_rate": 2e-05,
      "loss": 0.2041,
      "step": 978
    },
    {
      "epoch": 2.98,
      "learning_rate": 2e-05,
      "loss": 0.2066,
      "step": 979
    },
    {
      "epoch": 2.98,
      "learning_rate": 2e-05,
      "loss": 0.2002,
      "step": 980
    },
    {
      "epoch": 2.98,
      "learning_rate": 2e-05,
      "loss": 0.2035,
      "step": 981
    },
    {
      "epoch": 2.98,
      "learning_rate": 2e-05,
      "loss": 0.2208,
      "step": 982
    },
    {
      "epoch": 2.99,
      "learning_rate": 2e-05,
      "loss": 0.2002,
      "step": 983
    },
    {
      "epoch": 2.99,
      "learning_rate": 2e-05,
      "loss": 0.2027,
      "step": 984
    },
    {
      "epoch": 2.99,
      "learning_rate": 2e-05,
      "loss": 0.2093,
      "step": 985
    },
    {
      "epoch": 3.0,
      "learning_rate": 2e-05,
      "loss": 0.2103,
      "step": 986
    },
    {
      "epoch": 3.0,
      "learning_rate": 2e-05,
      "loss": 0.1926,
      "step": 987
    },
    {
      "epoch": 3.0,
      "step": 987,
      "total_flos": 1615206397378560.0,
      "train_loss": 0.2995875944754273,
      "train_runtime": 21762.911,
      "train_samples_per_second": 5.8,
      "train_steps_per_second": 0.045
    }
  ],
  "max_steps": 987,
  "num_train_epochs": 3,
  "total_flos": 1615206397378560.0,
  "trial_name": null,
  "trial_params": null
}