{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "global_step": 636,
  "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": 1.0788,
      "step": 1
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6275642631951835e-06,
      "loss": 0.8889,
      "step": 2
    },
    {
      "epoch": 0.01,
      "learning_rate": 7.3345158268416935e-06,
      "loss": 0.8858,
      "step": 3
    },
    {
      "epoch": 0.02,
      "learning_rate": 9.255128526390367e-06,
      "loss": 0.7224,
      "step": 4
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.0744871473609633e-05,
      "loss": 0.6558,
      "step": 5
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.1962080090036879e-05,
      "loss": 0.6163,
      "step": 6
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.299121531141887e-05,
      "loss": 0.5613,
      "step": 7
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.388269278958555e-05,
      "loss": 0.5535,
      "step": 8
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.4669031653683387e-05,
      "loss": 0.5335,
      "step": 9
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.537243573680482e-05,
      "loss": 0.5121,
      "step": 10
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.600874212937343e-05,
      "loss": 0.5067,
      "step": 11
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.6589644353232063e-05,
      "loss": 0.4877,
      "step": 12
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.712402259777778e-05,
      "loss": 0.4897,
      "step": 13
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.7618779574614054e-05,
      "loss": 0.4857,
      "step": 14
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.8079387300451327e-05,
      "loss": 0.4843,
      "step": 15
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.8510257052780734e-05,
      "loss": 0.4823,
      "step": 16
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.891499697130832e-05,
      "loss": 0.4675,
      "step": 17
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.929659591687857e-05,
      "loss": 0.4756,
      "step": 18
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.9657557553855117e-05,
      "loss": 0.4643,
      "step": 19
    },
    {
      "epoch": 0.09,
      "learning_rate": 2e-05,
      "loss": 0.4616,
      "step": 20
    },
    {
      "epoch": 0.1,
      "learning_rate": 2e-05,
      "loss": 0.4614,
      "step": 21
    },
    {
      "epoch": 0.1,
      "learning_rate": 2e-05,
      "loss": 0.4487,
      "step": 22
    },
    {
      "epoch": 0.11,
      "learning_rate": 2e-05,
      "loss": 0.4615,
      "step": 23
    },
    {
      "epoch": 0.11,
      "learning_rate": 2e-05,
      "loss": 0.4522,
      "step": 24
    },
    {
      "epoch": 0.12,
      "learning_rate": 2e-05,
      "loss": 0.4403,
      "step": 25
    },
    {
      "epoch": 0.12,
      "learning_rate": 2e-05,
      "loss": 0.4427,
      "step": 26
    },
    {
      "epoch": 0.13,
      "learning_rate": 2e-05,
      "loss": 0.4425,
      "step": 27
    },
    {
      "epoch": 0.13,
      "learning_rate": 2e-05,
      "loss": 0.4412,
      "step": 28
    },
    {
      "epoch": 0.14,
      "learning_rate": 2e-05,
      "loss": 0.4357,
      "step": 29
    },
    {
      "epoch": 0.14,
      "learning_rate": 2e-05,
      "loss": 0.4283,
      "step": 30
    },
    {
      "epoch": 0.15,
      "learning_rate": 2e-05,
      "loss": 0.437,
      "step": 31
    },
    {
      "epoch": 0.15,
      "learning_rate": 2e-05,
      "loss": 0.4321,
      "step": 32
    },
    {
      "epoch": 0.16,
      "learning_rate": 2e-05,
      "loss": 0.4313,
      "step": 33
    },
    {
      "epoch": 0.16,
      "learning_rate": 2e-05,
      "loss": 0.4116,
      "step": 34
    },
    {
      "epoch": 0.17,
      "learning_rate": 2e-05,
      "loss": 0.4124,
      "step": 35
    },
    {
      "epoch": 0.17,
      "learning_rate": 2e-05,
      "loss": 0.4226,
      "step": 36
    },
    {
      "epoch": 0.17,
      "learning_rate": 2e-05,
      "loss": 0.4152,
      "step": 37
    },
    {
      "epoch": 0.18,
      "learning_rate": 2e-05,
      "loss": 0.4099,
      "step": 38
    },
    {
      "epoch": 0.18,
      "learning_rate": 2e-05,
      "loss": 0.4086,
      "step": 39
    },
    {
      "epoch": 0.19,
      "learning_rate": 2e-05,
      "loss": 0.4149,
      "step": 40
    },
    {
      "epoch": 0.19,
      "learning_rate": 2e-05,
      "loss": 0.4114,
      "step": 41
    },
    {
      "epoch": 0.2,
      "learning_rate": 2e-05,
      "loss": 0.4103,
      "step": 42
    },
    {
      "epoch": 0.2,
      "learning_rate": 2e-05,
      "loss": 0.4067,
      "step": 43
    },
    {
      "epoch": 0.21,
      "learning_rate": 2e-05,
      "loss": 0.3959,
      "step": 44
    },
    {
      "epoch": 0.21,
      "learning_rate": 2e-05,
      "loss": 0.3945,
      "step": 45
    },
    {
      "epoch": 0.22,
      "learning_rate": 2e-05,
      "loss": 0.3925,
      "step": 46
    },
    {
      "epoch": 0.22,
      "learning_rate": 2e-05,
      "loss": 0.3902,
      "step": 47
    },
    {
      "epoch": 0.23,
      "learning_rate": 2e-05,
      "loss": 0.3824,
      "step": 48
    },
    {
      "epoch": 0.23,
      "learning_rate": 2e-05,
      "loss": 0.3905,
      "step": 49
    },
    {
      "epoch": 0.24,
      "learning_rate": 2e-05,
      "loss": 0.3984,
      "step": 50
    },
    {
      "epoch": 0.24,
      "learning_rate": 2e-05,
      "loss": 0.39,
      "step": 51
    },
    {
      "epoch": 0.25,
      "learning_rate": 2e-05,
      "loss": 0.3861,
      "step": 52
    },
    {
      "epoch": 0.25,
      "learning_rate": 2e-05,
      "loss": 0.3897,
      "step": 53
    },
    {
      "epoch": 0.25,
      "learning_rate": 2e-05,
      "loss": 0.3895,
      "step": 54
    },
    {
      "epoch": 0.26,
      "learning_rate": 2e-05,
      "loss": 0.3914,
      "step": 55
    },
    {
      "epoch": 0.26,
      "learning_rate": 2e-05,
      "loss": 0.3774,
      "step": 56
    },
    {
      "epoch": 0.27,
      "learning_rate": 2e-05,
      "loss": 0.3862,
      "step": 57
    },
    {
      "epoch": 0.27,
      "learning_rate": 2e-05,
      "loss": 0.3876,
      "step": 58
    },
    {
      "epoch": 0.28,
      "learning_rate": 2e-05,
      "loss": 0.3855,
      "step": 59
    },
    {
      "epoch": 0.28,
      "learning_rate": 2e-05,
      "loss": 0.3816,
      "step": 60
    },
    {
      "epoch": 0.29,
      "learning_rate": 2e-05,
      "loss": 0.3809,
      "step": 61
    },
    {
      "epoch": 0.29,
      "learning_rate": 2e-05,
      "loss": 0.3859,
      "step": 62
    },
    {
      "epoch": 0.3,
      "learning_rate": 2e-05,
      "loss": 0.3787,
      "step": 63
    },
    {
      "epoch": 0.3,
      "learning_rate": 2e-05,
      "loss": 0.3806,
      "step": 64
    },
    {
      "epoch": 0.31,
      "learning_rate": 2e-05,
      "loss": 0.3826,
      "step": 65
    },
    {
      "epoch": 0.31,
      "learning_rate": 2e-05,
      "loss": 0.3877,
      "step": 66
    },
    {
      "epoch": 0.32,
      "learning_rate": 2e-05,
      "loss": 0.3815,
      "step": 67
    },
    {
      "epoch": 0.32,
      "learning_rate": 2e-05,
      "loss": 0.3847,
      "step": 68
    },
    {
      "epoch": 0.33,
      "learning_rate": 2e-05,
      "loss": 0.3919,
      "step": 69
    },
    {
      "epoch": 0.33,
      "learning_rate": 2e-05,
      "loss": 0.3871,
      "step": 70
    },
    {
      "epoch": 0.33,
      "learning_rate": 2e-05,
      "loss": 0.3819,
      "step": 71
    },
    {
      "epoch": 0.34,
      "learning_rate": 2e-05,
      "loss": 0.3775,
      "step": 72
    },
    {
      "epoch": 0.34,
      "learning_rate": 2e-05,
      "loss": 0.3699,
      "step": 73
    },
    {
      "epoch": 0.35,
      "learning_rate": 2e-05,
      "loss": 0.3884,
      "step": 74
    },
    {
      "epoch": 0.35,
      "learning_rate": 2e-05,
      "loss": 0.3837,
      "step": 75
    },
    {
      "epoch": 0.36,
      "learning_rate": 2e-05,
      "loss": 0.3729,
      "step": 76
    },
    {
      "epoch": 0.36,
      "learning_rate": 2e-05,
      "loss": 0.3909,
      "step": 77
    },
    {
      "epoch": 0.37,
      "learning_rate": 2e-05,
      "loss": 0.3744,
      "step": 78
    },
    {
      "epoch": 0.37,
      "learning_rate": 2e-05,
      "loss": 0.3882,
      "step": 79
    },
    {
      "epoch": 0.38,
      "learning_rate": 2e-05,
      "loss": 0.3801,
      "step": 80
    },
    {
      "epoch": 0.38,
      "learning_rate": 2e-05,
      "loss": 0.378,
      "step": 81
    },
    {
      "epoch": 0.39,
      "learning_rate": 2e-05,
      "loss": 0.3767,
      "step": 82
    },
    {
      "epoch": 0.39,
      "learning_rate": 2e-05,
      "loss": 0.3831,
      "step": 83
    },
    {
      "epoch": 0.4,
      "learning_rate": 2e-05,
      "loss": 0.3779,
      "step": 84
    },
    {
      "epoch": 0.4,
      "learning_rate": 2e-05,
      "loss": 0.3773,
      "step": 85
    },
    {
      "epoch": 0.41,
      "learning_rate": 2e-05,
      "loss": 0.3669,
      "step": 86
    },
    {
      "epoch": 0.41,
      "learning_rate": 2e-05,
      "loss": 0.3867,
      "step": 87
    },
    {
      "epoch": 0.42,
      "learning_rate": 2e-05,
      "loss": 0.3704,
      "step": 88
    },
    {
      "epoch": 0.42,
      "learning_rate": 2e-05,
      "loss": 0.3636,
      "step": 89
    },
    {
      "epoch": 0.42,
      "learning_rate": 2e-05,
      "loss": 0.3768,
      "step": 90
    },
    {
      "epoch": 0.43,
      "learning_rate": 2e-05,
      "loss": 0.3796,
      "step": 91
    },
    {
      "epoch": 0.43,
      "learning_rate": 2e-05,
      "loss": 0.3836,
      "step": 92
    },
    {
      "epoch": 0.44,
      "learning_rate": 2e-05,
      "loss": 0.3722,
      "step": 93
    },
    {
      "epoch": 0.44,
      "learning_rate": 2e-05,
      "loss": 0.3687,
      "step": 94
    },
    {
      "epoch": 0.45,
      "learning_rate": 2e-05,
      "loss": 0.3805,
      "step": 95
    },
    {
      "epoch": 0.45,
      "learning_rate": 2e-05,
      "loss": 0.3725,
      "step": 96
    },
    {
      "epoch": 0.46,
      "learning_rate": 2e-05,
      "loss": 0.3629,
      "step": 97
    },
    {
      "epoch": 0.46,
      "learning_rate": 2e-05,
      "loss": 0.3781,
      "step": 98
    },
    {
      "epoch": 0.47,
      "learning_rate": 2e-05,
      "loss": 0.3616,
      "step": 99
    },
    {
      "epoch": 0.47,
      "learning_rate": 2e-05,
      "loss": 0.3665,
      "step": 100
    },
    {
      "epoch": 0.48,
      "learning_rate": 2e-05,
      "loss": 0.3761,
      "step": 101
    },
    {
      "epoch": 0.48,
      "learning_rate": 2e-05,
      "loss": 0.3798,
      "step": 102
    },
    {
      "epoch": 0.49,
      "learning_rate": 2e-05,
      "loss": 0.3663,
      "step": 103
    },
    {
      "epoch": 0.49,
      "learning_rate": 2e-05,
      "loss": 0.3715,
      "step": 104
    },
    {
      "epoch": 0.5,
      "learning_rate": 2e-05,
      "loss": 0.3738,
      "step": 105
    },
    {
      "epoch": 0.5,
      "learning_rate": 2e-05,
      "loss": 0.3656,
      "step": 106
    },
    {
      "epoch": 0.5,
      "learning_rate": 2e-05,
      "loss": 0.3706,
      "step": 107
    },
    {
      "epoch": 0.51,
      "learning_rate": 2e-05,
      "loss": 0.3716,
      "step": 108
    },
    {
      "epoch": 0.51,
      "learning_rate": 2e-05,
      "loss": 0.3739,
      "step": 109
    },
    {
      "epoch": 0.52,
      "learning_rate": 2e-05,
      "loss": 0.368,
      "step": 110
    },
    {
      "epoch": 0.52,
      "learning_rate": 2e-05,
      "loss": 0.3802,
      "step": 111
    },
    {
      "epoch": 0.53,
      "learning_rate": 2e-05,
      "loss": 0.3614,
      "step": 112
    },
    {
      "epoch": 0.53,
      "learning_rate": 2e-05,
      "loss": 0.3655,
      "step": 113
    },
    {
      "epoch": 0.54,
      "learning_rate": 2e-05,
      "loss": 0.367,
      "step": 114
    },
    {
      "epoch": 0.54,
      "learning_rate": 2e-05,
      "loss": 0.3733,
      "step": 115
    },
    {
      "epoch": 0.55,
      "learning_rate": 2e-05,
      "loss": 0.3545,
      "step": 116
    },
    {
      "epoch": 0.55,
      "learning_rate": 2e-05,
      "loss": 0.3706,
      "step": 117
    },
    {
      "epoch": 0.56,
      "learning_rate": 2e-05,
      "loss": 0.3766,
      "step": 118
    },
    {
      "epoch": 0.56,
      "learning_rate": 2e-05,
      "loss": 0.3676,
      "step": 119
    },
    {
      "epoch": 0.57,
      "learning_rate": 2e-05,
      "loss": 0.3635,
      "step": 120
    },
    {
      "epoch": 0.57,
      "learning_rate": 2e-05,
      "loss": 0.3633,
      "step": 121
    },
    {
      "epoch": 0.58,
      "learning_rate": 2e-05,
      "loss": 0.3662,
      "step": 122
    },
    {
      "epoch": 0.58,
      "learning_rate": 2e-05,
      "loss": 0.3604,
      "step": 123
    },
    {
      "epoch": 0.58,
      "learning_rate": 2e-05,
      "loss": 0.3675,
      "step": 124
    },
    {
      "epoch": 0.59,
      "learning_rate": 2e-05,
      "loss": 0.3763,
      "step": 125
    },
    {
      "epoch": 0.59,
      "learning_rate": 2e-05,
      "loss": 0.3621,
      "step": 126
    },
    {
      "epoch": 0.6,
      "learning_rate": 2e-05,
      "loss": 0.3608,
      "step": 127
    },
    {
      "epoch": 0.6,
      "learning_rate": 2e-05,
      "loss": 0.367,
      "step": 128
    },
    {
      "epoch": 0.61,
      "learning_rate": 2e-05,
      "loss": 0.3735,
      "step": 129
    },
    {
      "epoch": 0.61,
      "learning_rate": 2e-05,
      "loss": 0.3612,
      "step": 130
    },
    {
      "epoch": 0.62,
      "learning_rate": 2e-05,
      "loss": 0.3711,
      "step": 131
    },
    {
      "epoch": 0.62,
      "learning_rate": 2e-05,
      "loss": 0.3744,
      "step": 132
    },
    {
      "epoch": 0.63,
      "learning_rate": 2e-05,
      "loss": 0.3746,
      "step": 133
    },
    {
      "epoch": 0.63,
      "learning_rate": 2e-05,
      "loss": 0.3614,
      "step": 134
    },
    {
      "epoch": 0.64,
      "learning_rate": 2e-05,
      "loss": 0.3679,
      "step": 135
    },
    {
      "epoch": 0.64,
      "learning_rate": 2e-05,
      "loss": 0.3544,
      "step": 136
    },
    {
      "epoch": 0.65,
      "learning_rate": 2e-05,
      "loss": 0.3581,
      "step": 137
    },
    {
      "epoch": 0.65,
      "learning_rate": 2e-05,
      "loss": 0.3675,
      "step": 138
    },
    {
      "epoch": 0.66,
      "learning_rate": 2e-05,
      "loss": 0.3617,
      "step": 139
    },
    {
      "epoch": 0.66,
      "learning_rate": 2e-05,
      "loss": 0.3685,
      "step": 140
    },
    {
      "epoch": 0.67,
      "learning_rate": 2e-05,
      "loss": 0.3548,
      "step": 141
    },
    {
      "epoch": 0.67,
      "learning_rate": 2e-05,
      "loss": 0.361,
      "step": 142
    },
    {
      "epoch": 0.67,
      "learning_rate": 2e-05,
      "loss": 0.3649,
      "step": 143
    },
    {
      "epoch": 0.68,
      "learning_rate": 2e-05,
      "loss": 0.3537,
      "step": 144
    },
    {
      "epoch": 0.68,
      "learning_rate": 2e-05,
      "loss": 0.3734,
      "step": 145
    },
    {
      "epoch": 0.69,
      "learning_rate": 2e-05,
      "loss": 0.3519,
      "step": 146
    },
    {
      "epoch": 0.69,
      "learning_rate": 2e-05,
      "loss": 0.3494,
      "step": 147
    },
    {
      "epoch": 0.7,
      "learning_rate": 2e-05,
      "loss": 0.3689,
      "step": 148
    },
    {
      "epoch": 0.7,
      "learning_rate": 2e-05,
      "loss": 0.3612,
      "step": 149
    },
    {
      "epoch": 0.71,
      "learning_rate": 2e-05,
      "loss": 0.3585,
      "step": 150
    },
    {
      "epoch": 0.71,
      "learning_rate": 2e-05,
      "loss": 0.3526,
      "step": 151
    },
    {
      "epoch": 0.72,
      "learning_rate": 2e-05,
      "loss": 0.3631,
      "step": 152
    },
    {
      "epoch": 0.72,
      "learning_rate": 2e-05,
      "loss": 0.3564,
      "step": 153
    },
    {
      "epoch": 0.73,
      "learning_rate": 2e-05,
      "loss": 0.3524,
      "step": 154
    },
    {
      "epoch": 0.73,
      "learning_rate": 2e-05,
      "loss": 0.3632,
      "step": 155
    },
    {
      "epoch": 0.74,
      "learning_rate": 2e-05,
      "loss": 0.3556,
      "step": 156
    },
    {
      "epoch": 0.74,
      "learning_rate": 2e-05,
      "loss": 0.3499,
      "step": 157
    },
    {
      "epoch": 0.75,
      "learning_rate": 2e-05,
      "loss": 0.3506,
      "step": 158
    },
    {
      "epoch": 0.75,
      "learning_rate": 2e-05,
      "loss": 0.3627,
      "step": 159
    },
    {
      "epoch": 0.75,
      "learning_rate": 2e-05,
      "loss": 0.3608,
      "step": 160
    },
    {
      "epoch": 0.76,
      "learning_rate": 2e-05,
      "loss": 0.3492,
      "step": 161
    },
    {
      "epoch": 0.76,
      "learning_rate": 2e-05,
      "loss": 0.3524,
      "step": 162
    },
    {
      "epoch": 0.77,
      "learning_rate": 2e-05,
      "loss": 0.3566,
      "step": 163
    },
    {
      "epoch": 0.77,
      "learning_rate": 2e-05,
      "loss": 0.3543,
      "step": 164
    },
    {
      "epoch": 0.78,
      "learning_rate": 2e-05,
      "loss": 0.3526,
      "step": 165
    },
    {
      "epoch": 0.78,
      "learning_rate": 2e-05,
      "loss": 0.3593,
      "step": 166
    },
    {
      "epoch": 0.79,
      "learning_rate": 2e-05,
      "loss": 0.3563,
      "step": 167
    },
    {
      "epoch": 0.79,
      "learning_rate": 2e-05,
      "loss": 0.3508,
      "step": 168
    },
    {
      "epoch": 0.8,
      "learning_rate": 2e-05,
      "loss": 0.3567,
      "step": 169
    },
    {
      "epoch": 0.8,
      "learning_rate": 2e-05,
      "loss": 0.363,
      "step": 170
    },
    {
      "epoch": 0.81,
      "learning_rate": 2e-05,
      "loss": 0.3524,
      "step": 171
    },
    {
      "epoch": 0.81,
      "learning_rate": 2e-05,
      "loss": 0.3583,
      "step": 172
    },
    {
      "epoch": 0.82,
      "learning_rate": 2e-05,
      "loss": 0.3565,
      "step": 173
    },
    {
      "epoch": 0.82,
      "learning_rate": 2e-05,
      "loss": 0.3393,
      "step": 174
    },
    {
      "epoch": 0.83,
      "learning_rate": 2e-05,
      "loss": 0.3525,
      "step": 175
    },
    {
      "epoch": 0.83,
      "learning_rate": 2e-05,
      "loss": 0.3459,
      "step": 176
    },
    {
      "epoch": 0.83,
      "learning_rate": 2e-05,
      "loss": 0.3581,
      "step": 177
    },
    {
      "epoch": 0.84,
      "learning_rate": 2e-05,
      "loss": 0.3658,
      "step": 178
    },
    {
      "epoch": 0.84,
      "learning_rate": 2e-05,
      "loss": 0.3557,
      "step": 179
    },
    {
      "epoch": 0.85,
      "learning_rate": 2e-05,
      "loss": 0.3532,
      "step": 180
    },
    {
      "epoch": 0.85,
      "learning_rate": 2e-05,
      "loss": 0.3553,
      "step": 181
    },
    {
      "epoch": 0.86,
      "learning_rate": 2e-05,
      "loss": 0.345,
      "step": 182
    },
    {
      "epoch": 0.86,
      "learning_rate": 2e-05,
      "loss": 0.3659,
      "step": 183
    },
    {
      "epoch": 0.87,
      "learning_rate": 2e-05,
      "loss": 0.3468,
      "step": 184
    },
    {
      "epoch": 0.87,
      "learning_rate": 2e-05,
      "loss": 0.3455,
      "step": 185
    },
    {
      "epoch": 0.88,
      "learning_rate": 2e-05,
      "loss": 0.354,
      "step": 186
    },
    {
      "epoch": 0.88,
      "learning_rate": 2e-05,
      "loss": 0.357,
      "step": 187
    },
    {
      "epoch": 0.89,
      "learning_rate": 2e-05,
      "loss": 0.3583,
      "step": 188
    },
    {
      "epoch": 0.89,
      "learning_rate": 2e-05,
      "loss": 0.3606,
      "step": 189
    },
    {
      "epoch": 0.9,
      "learning_rate": 2e-05,
      "loss": 0.351,
      "step": 190
    },
    {
      "epoch": 0.9,
      "learning_rate": 2e-05,
      "loss": 0.3507,
      "step": 191
    },
    {
      "epoch": 0.91,
      "learning_rate": 2e-05,
      "loss": 0.3577,
      "step": 192
    },
    {
      "epoch": 0.91,
      "learning_rate": 2e-05,
      "loss": 0.3561,
      "step": 193
    },
    {
      "epoch": 0.92,
      "learning_rate": 2e-05,
      "loss": 0.3442,
      "step": 194
    },
    {
      "epoch": 0.92,
      "learning_rate": 2e-05,
      "loss": 0.3533,
      "step": 195
    },
    {
      "epoch": 0.92,
      "learning_rate": 2e-05,
      "loss": 0.347,
      "step": 196
    },
    {
      "epoch": 0.93,
      "learning_rate": 2e-05,
      "loss": 0.3537,
      "step": 197
    },
    {
      "epoch": 0.93,
      "learning_rate": 2e-05,
      "loss": 0.3622,
      "step": 198
    },
    {
      "epoch": 0.94,
      "learning_rate": 2e-05,
      "loss": 0.3574,
      "step": 199
    },
    {
      "epoch": 0.94,
      "learning_rate": 2e-05,
      "loss": 0.3509,
      "step": 200
    },
    {
      "epoch": 0.95,
      "learning_rate": 2e-05,
      "loss": 0.3557,
      "step": 201
    },
    {
      "epoch": 0.95,
      "learning_rate": 2e-05,
      "loss": 0.3527,
      "step": 202
    },
    {
      "epoch": 0.96,
      "learning_rate": 2e-05,
      "loss": 0.3578,
      "step": 203
    },
    {
      "epoch": 0.96,
      "learning_rate": 2e-05,
      "loss": 0.3435,
      "step": 204
    },
    {
      "epoch": 0.97,
      "learning_rate": 2e-05,
      "loss": 0.3482,
      "step": 205
    },
    {
      "epoch": 0.97,
      "learning_rate": 2e-05,
      "loss": 0.3527,
      "step": 206
    },
    {
      "epoch": 0.98,
      "learning_rate": 2e-05,
      "loss": 0.365,
      "step": 207
    },
    {
      "epoch": 0.98,
      "learning_rate": 2e-05,
      "loss": 0.3574,
      "step": 208
    },
    {
      "epoch": 0.99,
      "learning_rate": 2e-05,
      "loss": 0.3518,
      "step": 209
    },
    {
      "epoch": 0.99,
      "learning_rate": 2e-05,
      "loss": 0.346,
      "step": 210
    },
    {
      "epoch": 1.0,
      "learning_rate": 2e-05,
      "loss": 0.344,
      "step": 211
    },
    {
      "epoch": 1.0,
      "learning_rate": 2e-05,
      "loss": 0.3554,
      "step": 212
    },
    {
      "epoch": 1.0,
      "learning_rate": 2e-05,
      "loss": 0.3281,
      "step": 213
    },
    {
      "epoch": 1.01,
      "learning_rate": 2e-05,
      "loss": 0.323,
      "step": 214
    },
    {
      "epoch": 1.01,
      "learning_rate": 2e-05,
      "loss": 0.3289,
      "step": 215
    },
    {
      "epoch": 1.02,
      "learning_rate": 2e-05,
      "loss": 0.3312,
      "step": 216
    },
    {
      "epoch": 1.02,
      "learning_rate": 2e-05,
      "loss": 0.3197,
      "step": 217
    },
    {
      "epoch": 1.03,
      "learning_rate": 2e-05,
      "loss": 0.3069,
      "step": 218
    },
    {
      "epoch": 1.03,
      "learning_rate": 2e-05,
      "loss": 0.3168,
      "step": 219
    },
    {
      "epoch": 1.04,
      "learning_rate": 2e-05,
      "loss": 0.3288,
      "step": 220
    },
    {
      "epoch": 1.04,
      "learning_rate": 2e-05,
      "loss": 0.306,
      "step": 221
    },
    {
      "epoch": 1.05,
      "learning_rate": 2e-05,
      "loss": 0.3213,
      "step": 222
    },
    {
      "epoch": 1.05,
      "learning_rate": 2e-05,
      "loss": 0.3223,
      "step": 223
    },
    {
      "epoch": 1.06,
      "learning_rate": 2e-05,
      "loss": 0.3164,
      "step": 224
    },
    {
      "epoch": 1.06,
      "learning_rate": 2e-05,
      "loss": 0.3209,
      "step": 225
    },
    {
      "epoch": 1.07,
      "learning_rate": 2e-05,
      "loss": 0.3128,
      "step": 226
    },
    {
      "epoch": 1.07,
      "learning_rate": 2e-05,
      "loss": 0.3198,
      "step": 227
    },
    {
      "epoch": 1.08,
      "learning_rate": 2e-05,
      "loss": 0.3147,
      "step": 228
    },
    {
      "epoch": 1.08,
      "learning_rate": 2e-05,
      "loss": 0.3092,
      "step": 229
    },
    {
      "epoch": 1.08,
      "learning_rate": 2e-05,
      "loss": 0.3145,
      "step": 230
    },
    {
      "epoch": 1.09,
      "learning_rate": 2e-05,
      "loss": 0.313,
      "step": 231
    },
    {
      "epoch": 1.09,
      "learning_rate": 2e-05,
      "loss": 0.3166,
      "step": 232
    },
    {
      "epoch": 1.1,
      "learning_rate": 2e-05,
      "loss": 0.3237,
      "step": 233
    },
    {
      "epoch": 1.1,
      "learning_rate": 2e-05,
      "loss": 0.3045,
      "step": 234
    },
    {
      "epoch": 1.11,
      "learning_rate": 2e-05,
      "loss": 0.3285,
      "step": 235
    },
    {
      "epoch": 1.11,
      "learning_rate": 2e-05,
      "loss": 0.3135,
      "step": 236
    },
    {
      "epoch": 1.12,
      "learning_rate": 2e-05,
      "loss": 0.318,
      "step": 237
    },
    {
      "epoch": 1.12,
      "learning_rate": 2e-05,
      "loss": 0.319,
      "step": 238
    },
    {
      "epoch": 1.13,
      "learning_rate": 2e-05,
      "loss": 0.3189,
      "step": 239
    },
    {
      "epoch": 1.13,
      "learning_rate": 2e-05,
      "loss": 0.3235,
      "step": 240
    },
    {
      "epoch": 1.14,
      "learning_rate": 2e-05,
      "loss": 0.3146,
      "step": 241
    },
    {
      "epoch": 1.14,
      "learning_rate": 2e-05,
      "loss": 0.317,
      "step": 242
    },
    {
      "epoch": 1.15,
      "learning_rate": 2e-05,
      "loss": 0.3328,
      "step": 243
    },
    {
      "epoch": 1.15,
      "learning_rate": 2e-05,
      "loss": 0.3186,
      "step": 244
    },
    {
      "epoch": 1.16,
      "learning_rate": 2e-05,
      "loss": 0.3323,
      "step": 245
    },
    {
      "epoch": 1.16,
      "learning_rate": 2e-05,
      "loss": 0.3173,
      "step": 246
    },
    {
      "epoch": 1.17,
      "learning_rate": 2e-05,
      "loss": 0.3235,
      "step": 247
    },
    {
      "epoch": 1.17,
      "learning_rate": 2e-05,
      "loss": 0.3224,
      "step": 248
    },
    {
      "epoch": 1.17,
      "learning_rate": 2e-05,
      "loss": 0.3207,
      "step": 249
    },
    {
      "epoch": 1.18,
      "learning_rate": 2e-05,
      "loss": 0.3319,
      "step": 250
    },
    {
      "epoch": 1.18,
      "learning_rate": 2e-05,
      "loss": 0.3213,
      "step": 251
    },
    {
      "epoch": 1.19,
      "learning_rate": 2e-05,
      "loss": 0.3278,
      "step": 252
    },
    {
      "epoch": 1.19,
      "learning_rate": 2e-05,
      "loss": 0.3246,
      "step": 253
    },
    {
      "epoch": 1.2,
      "learning_rate": 2e-05,
      "loss": 0.3262,
      "step": 254
    },
    {
      "epoch": 1.2,
      "learning_rate": 2e-05,
      "loss": 0.3028,
      "step": 255
    },
    {
      "epoch": 1.21,
      "learning_rate": 2e-05,
      "loss": 0.3115,
      "step": 256
    },
    {
      "epoch": 1.21,
      "learning_rate": 2e-05,
      "loss": 0.307,
      "step": 257
    },
    {
      "epoch": 1.22,
      "learning_rate": 2e-05,
      "loss": 0.316,
      "step": 258
    },
    {
      "epoch": 1.22,
      "learning_rate": 2e-05,
      "loss": 0.323,
      "step": 259
    },
    {
      "epoch": 1.23,
      "learning_rate": 2e-05,
      "loss": 0.3152,
      "step": 260
    },
    {
      "epoch": 1.23,
      "learning_rate": 2e-05,
      "loss": 0.3179,
      "step": 261
    },
    {
      "epoch": 1.24,
      "learning_rate": 2e-05,
      "loss": 0.3131,
      "step": 262
    },
    {
      "epoch": 1.24,
      "learning_rate": 2e-05,
      "loss": 0.3132,
      "step": 263
    },
    {
      "epoch": 1.25,
      "learning_rate": 2e-05,
      "loss": 0.3049,
      "step": 264
    },
    {
      "epoch": 1.25,
      "learning_rate": 2e-05,
      "loss": 0.3215,
      "step": 265
    },
    {
      "epoch": 1.25,
      "learning_rate": 2e-05,
      "loss": 0.311,
      "step": 266
    },
    {
      "epoch": 1.26,
      "learning_rate": 2e-05,
      "loss": 0.3166,
      "step": 267
    },
    {
      "epoch": 1.26,
      "learning_rate": 2e-05,
      "loss": 0.3067,
      "step": 268
    },
    {
      "epoch": 1.27,
      "learning_rate": 2e-05,
      "loss": 0.3145,
      "step": 269
    },
    {
      "epoch": 1.27,
      "learning_rate": 2e-05,
      "loss": 0.3108,
      "step": 270
    },
    {
      "epoch": 1.28,
      "learning_rate": 2e-05,
      "loss": 0.3166,
      "step": 271
    },
    {
      "epoch": 1.28,
      "learning_rate": 2e-05,
      "loss": 0.3299,
      "step": 272
    },
    {
      "epoch": 1.29,
      "learning_rate": 2e-05,
      "loss": 0.3211,
      "step": 273
    },
    {
      "epoch": 1.29,
      "learning_rate": 2e-05,
      "loss": 0.3038,
      "step": 274
    },
    {
      "epoch": 1.3,
      "learning_rate": 2e-05,
      "loss": 0.3247,
      "step": 275
    },
    {
      "epoch": 1.3,
      "learning_rate": 2e-05,
      "loss": 0.3243,
      "step": 276
    },
    {
      "epoch": 1.31,
      "learning_rate": 2e-05,
      "loss": 0.3214,
      "step": 277
    },
    {
      "epoch": 1.31,
      "learning_rate": 2e-05,
      "loss": 0.3117,
      "step": 278
    },
    {
      "epoch": 1.32,
      "learning_rate": 2e-05,
      "loss": 0.3224,
      "step": 279
    },
    {
      "epoch": 1.32,
      "learning_rate": 2e-05,
      "loss": 0.3188,
      "step": 280
    },
    {
      "epoch": 1.33,
      "learning_rate": 2e-05,
      "loss": 0.3212,
      "step": 281
    },
    {
      "epoch": 1.33,
      "learning_rate": 2e-05,
      "loss": 0.3211,
      "step": 282
    },
    {
      "epoch": 1.33,
      "learning_rate": 2e-05,
      "loss": 0.3251,
      "step": 283
    },
    {
      "epoch": 1.34,
      "learning_rate": 2e-05,
      "loss": 0.3105,
      "step": 284
    },
    {
      "epoch": 1.34,
      "learning_rate": 2e-05,
      "loss": 0.3176,
      "step": 285
    },
    {
      "epoch": 1.35,
      "learning_rate": 2e-05,
      "loss": 0.3199,
      "step": 286
    },
    {
      "epoch": 1.35,
      "learning_rate": 2e-05,
      "loss": 0.3252,
      "step": 287
    },
    {
      "epoch": 1.36,
      "learning_rate": 2e-05,
      "loss": 0.3159,
      "step": 288
    },
    {
      "epoch": 1.36,
      "learning_rate": 2e-05,
      "loss": 0.3207,
      "step": 289
    },
    {
      "epoch": 1.37,
      "learning_rate": 2e-05,
      "loss": 0.3159,
      "step": 290
    },
    {
      "epoch": 1.37,
      "learning_rate": 2e-05,
      "loss": 0.3285,
      "step": 291
    },
    {
      "epoch": 1.38,
      "learning_rate": 2e-05,
      "loss": 0.3226,
      "step": 292
    },
    {
      "epoch": 1.38,
      "learning_rate": 2e-05,
      "loss": 0.3202,
      "step": 293
    },
    {
      "epoch": 1.39,
      "learning_rate": 2e-05,
      "loss": 0.3107,
      "step": 294
    },
    {
      "epoch": 1.39,
      "learning_rate": 2e-05,
      "loss": 0.3238,
      "step": 295
    },
    {
      "epoch": 1.4,
      "learning_rate": 2e-05,
      "loss": 0.3218,
      "step": 296
    },
    {
      "epoch": 1.4,
      "learning_rate": 2e-05,
      "loss": 0.3128,
      "step": 297
    },
    {
      "epoch": 1.41,
      "learning_rate": 2e-05,
      "loss": 0.3147,
      "step": 298
    },
    {
      "epoch": 1.41,
      "learning_rate": 2e-05,
      "loss": 0.3191,
      "step": 299
    },
    {
      "epoch": 1.42,
      "learning_rate": 2e-05,
      "loss": 0.3184,
      "step": 300
    },
    {
      "epoch": 1.42,
      "learning_rate": 2e-05,
      "loss": 0.3172,
      "step": 301
    },
    {
      "epoch": 1.42,
      "learning_rate": 2e-05,
      "loss": 0.317,
      "step": 302
    },
    {
      "epoch": 1.43,
      "learning_rate": 2e-05,
      "loss": 0.3179,
      "step": 303
    },
    {
      "epoch": 1.43,
      "learning_rate": 2e-05,
      "loss": 0.3162,
      "step": 304
    },
    {
      "epoch": 1.44,
      "learning_rate": 2e-05,
      "loss": 0.3082,
      "step": 305
    },
    {
      "epoch": 1.44,
      "learning_rate": 2e-05,
      "loss": 0.3074,
      "step": 306
    },
    {
      "epoch": 1.45,
      "learning_rate": 2e-05,
      "loss": 0.3141,
      "step": 307
    },
    {
      "epoch": 1.45,
      "learning_rate": 2e-05,
      "loss": 0.3092,
      "step": 308
    },
    {
      "epoch": 1.46,
      "learning_rate": 2e-05,
      "loss": 0.3188,
      "step": 309
    },
    {
      "epoch": 1.46,
      "learning_rate": 2e-05,
      "loss": 0.3153,
      "step": 310
    },
    {
      "epoch": 1.47,
      "learning_rate": 2e-05,
      "loss": 0.3117,
      "step": 311
    },
    {
      "epoch": 1.47,
      "learning_rate": 2e-05,
      "loss": 0.3177,
      "step": 312
    },
    {
      "epoch": 1.48,
      "learning_rate": 2e-05,
      "loss": 0.3136,
      "step": 313
    },
    {
      "epoch": 1.48,
      "learning_rate": 2e-05,
      "loss": 0.3207,
      "step": 314
    },
    {
      "epoch": 1.49,
      "learning_rate": 2e-05,
      "loss": 0.3149,
      "step": 315
    },
    {
      "epoch": 1.49,
      "learning_rate": 2e-05,
      "loss": 0.3126,
      "step": 316
    },
    {
      "epoch": 1.5,
      "learning_rate": 2e-05,
      "loss": 0.3108,
      "step": 317
    },
    {
      "epoch": 1.5,
      "learning_rate": 2e-05,
      "loss": 0.3099,
      "step": 318
    },
    {
      "epoch": 1.5,
      "learning_rate": 2e-05,
      "loss": 0.3049,
      "step": 319
    },
    {
      "epoch": 1.51,
      "learning_rate": 2e-05,
      "loss": 0.3119,
      "step": 320
    },
    {
      "epoch": 1.51,
      "learning_rate": 2e-05,
      "loss": 0.3162,
      "step": 321
    },
    {
      "epoch": 1.52,
      "learning_rate": 2e-05,
      "loss": 0.3103,
      "step": 322
    },
    {
      "epoch": 1.52,
      "learning_rate": 2e-05,
      "loss": 0.3088,
      "step": 323
    },
    {
      "epoch": 1.53,
      "learning_rate": 2e-05,
      "loss": 0.3205,
      "step": 324
    },
    {
      "epoch": 1.53,
      "learning_rate": 2e-05,
      "loss": 0.311,
      "step": 325
    },
    {
      "epoch": 1.54,
      "learning_rate": 2e-05,
      "loss": 0.3172,
      "step": 326
    },
    {
      "epoch": 1.54,
      "learning_rate": 2e-05,
      "loss": 0.3171,
      "step": 327
    },
    {
      "epoch": 1.55,
      "learning_rate": 2e-05,
      "loss": 0.3187,
      "step": 328
    },
    {
      "epoch": 1.55,
      "learning_rate": 2e-05,
      "loss": 0.3098,
      "step": 329
    },
    {
      "epoch": 1.56,
      "learning_rate": 2e-05,
      "loss": 0.3198,
      "step": 330
    },
    {
      "epoch": 1.56,
      "learning_rate": 2e-05,
      "loss": 0.3178,
      "step": 331
    },
    {
      "epoch": 1.57,
      "learning_rate": 2e-05,
      "loss": 0.3163,
      "step": 332
    },
    {
      "epoch": 1.57,
      "learning_rate": 2e-05,
      "loss": 0.3167,
      "step": 333
    },
    {
      "epoch": 1.58,
      "learning_rate": 2e-05,
      "loss": 0.3085,
      "step": 334
    },
    {
      "epoch": 1.58,
      "learning_rate": 2e-05,
      "loss": 0.3088,
      "step": 335
    },
    {
      "epoch": 1.58,
      "learning_rate": 2e-05,
      "loss": 0.3152,
      "step": 336
    },
    {
      "epoch": 1.59,
      "learning_rate": 2e-05,
      "loss": 0.3177,
      "step": 337
    },
    {
      "epoch": 1.59,
      "learning_rate": 2e-05,
      "loss": 0.3275,
      "step": 338
    },
    {
      "epoch": 1.6,
      "learning_rate": 2e-05,
      "loss": 0.3153,
      "step": 339
    },
    {
      "epoch": 1.6,
      "learning_rate": 2e-05,
      "loss": 0.3127,
      "step": 340
    },
    {
      "epoch": 1.61,
      "learning_rate": 2e-05,
      "loss": 0.3114,
      "step": 341
    },
    {
      "epoch": 1.61,
      "learning_rate": 2e-05,
      "loss": 0.3,
      "step": 342
    },
    {
      "epoch": 1.62,
      "learning_rate": 2e-05,
      "loss": 0.3044,
      "step": 343
    },
    {
      "epoch": 1.62,
      "learning_rate": 2e-05,
      "loss": 0.3179,
      "step": 344
    },
    {
      "epoch": 1.63,
      "learning_rate": 2e-05,
      "loss": 0.309,
      "step": 345
    },
    {
      "epoch": 1.63,
      "learning_rate": 2e-05,
      "loss": 0.3139,
      "step": 346
    },
    {
      "epoch": 1.64,
      "learning_rate": 2e-05,
      "loss": 0.3039,
      "step": 347
    },
    {
      "epoch": 1.64,
      "learning_rate": 2e-05,
      "loss": 0.3106,
      "step": 348
    },
    {
      "epoch": 1.65,
      "learning_rate": 2e-05,
      "loss": 0.3048,
      "step": 349
    },
    {
      "epoch": 1.65,
      "learning_rate": 2e-05,
      "loss": 0.3133,
      "step": 350
    },
    {
      "epoch": 1.66,
      "learning_rate": 2e-05,
      "loss": 0.3175,
      "step": 351
    },
    {
      "epoch": 1.66,
      "learning_rate": 2e-05,
      "loss": 0.3083,
      "step": 352
    },
    {
      "epoch": 1.67,
      "learning_rate": 2e-05,
      "loss": 0.3083,
      "step": 353
    },
    {
      "epoch": 1.67,
      "learning_rate": 2e-05,
      "loss": 0.3104,
      "step": 354
    },
    {
      "epoch": 1.67,
      "learning_rate": 2e-05,
      "loss": 0.305,
      "step": 355
    },
    {
      "epoch": 1.68,
      "learning_rate": 2e-05,
      "loss": 0.3123,
      "step": 356
    },
    {
      "epoch": 1.68,
      "learning_rate": 2e-05,
      "loss": 0.3121,
      "step": 357
    },
    {
      "epoch": 1.69,
      "learning_rate": 2e-05,
      "loss": 0.3058,
      "step": 358
    },
    {
      "epoch": 1.69,
      "learning_rate": 2e-05,
      "loss": 0.3105,
      "step": 359
    },
    {
      "epoch": 1.7,
      "learning_rate": 2e-05,
      "loss": 0.3144,
      "step": 360
    },
    {
      "epoch": 1.7,
      "learning_rate": 2e-05,
      "loss": 0.3171,
      "step": 361
    },
    {
      "epoch": 1.71,
      "learning_rate": 2e-05,
      "loss": 0.2964,
      "step": 362
    },
    {
      "epoch": 1.71,
      "learning_rate": 2e-05,
      "loss": 0.3079,
      "step": 363
    },
    {
      "epoch": 1.72,
      "learning_rate": 2e-05,
      "loss": 0.3202,
      "step": 364
    },
    {
      "epoch": 1.72,
      "learning_rate": 2e-05,
      "loss": 0.3096,
      "step": 365
    },
    {
      "epoch": 1.73,
      "learning_rate": 2e-05,
      "loss": 0.3171,
      "step": 366
    },
    {
      "epoch": 1.73,
      "learning_rate": 2e-05,
      "loss": 0.3099,
      "step": 367
    },
    {
      "epoch": 1.74,
      "learning_rate": 2e-05,
      "loss": 0.2955,
      "step": 368
    },
    {
      "epoch": 1.74,
      "learning_rate": 2e-05,
      "loss": 0.3068,
      "step": 369
    },
    {
      "epoch": 1.75,
      "learning_rate": 2e-05,
      "loss": 0.3081,
      "step": 370
    },
    {
      "epoch": 1.75,
      "learning_rate": 2e-05,
      "loss": 0.3113,
      "step": 371
    },
    {
      "epoch": 1.75,
      "learning_rate": 2e-05,
      "loss": 0.3099,
      "step": 372
    },
    {
      "epoch": 1.76,
      "learning_rate": 2e-05,
      "loss": 0.3037,
      "step": 373
    },
    {
      "epoch": 1.76,
      "learning_rate": 2e-05,
      "loss": 0.3012,
      "step": 374
    },
    {
      "epoch": 1.77,
      "learning_rate": 2e-05,
      "loss": 0.3086,
      "step": 375
    },
    {
      "epoch": 1.77,
      "learning_rate": 2e-05,
      "loss": 0.3085,
      "step": 376
    },
    {
      "epoch": 1.78,
      "learning_rate": 2e-05,
      "loss": 0.3037,
      "step": 377
    },
    {
      "epoch": 1.78,
      "learning_rate": 2e-05,
      "loss": 0.2997,
      "step": 378
    },
    {
      "epoch": 1.79,
      "learning_rate": 2e-05,
      "loss": 0.3061,
      "step": 379
    },
    {
      "epoch": 1.79,
      "learning_rate": 2e-05,
      "loss": 0.3172,
      "step": 380
    },
    {
      "epoch": 1.8,
      "learning_rate": 2e-05,
      "loss": 0.307,
      "step": 381
    },
    {
      "epoch": 1.8,
      "learning_rate": 2e-05,
      "loss": 0.3059,
      "step": 382
    },
    {
      "epoch": 1.81,
      "learning_rate": 2e-05,
      "loss": 0.3027,
      "step": 383
    },
    {
      "epoch": 1.81,
      "learning_rate": 2e-05,
      "loss": 0.3148,
      "step": 384
    },
    {
      "epoch": 1.82,
      "learning_rate": 2e-05,
      "loss": 0.3013,
      "step": 385
    },
    {
      "epoch": 1.82,
      "learning_rate": 2e-05,
      "loss": 0.3109,
      "step": 386
    },
    {
      "epoch": 1.83,
      "learning_rate": 2e-05,
      "loss": 0.3034,
      "step": 387
    },
    {
      "epoch": 1.83,
      "learning_rate": 2e-05,
      "loss": 0.2941,
      "step": 388
    },
    {
      "epoch": 1.83,
      "learning_rate": 2e-05,
      "loss": 0.3071,
      "step": 389
    },
    {
      "epoch": 1.84,
      "learning_rate": 2e-05,
      "loss": 0.3026,
      "step": 390
    },
    {
      "epoch": 1.84,
      "learning_rate": 2e-05,
      "loss": 0.3118,
      "step": 391
    },
    {
      "epoch": 1.85,
      "learning_rate": 2e-05,
      "loss": 0.3158,
      "step": 392
    },
    {
      "epoch": 1.85,
      "learning_rate": 2e-05,
      "loss": 0.3056,
      "step": 393
    },
    {
      "epoch": 1.86,
      "learning_rate": 2e-05,
      "loss": 0.3016,
      "step": 394
    },
    {
      "epoch": 1.86,
      "learning_rate": 2e-05,
      "loss": 0.3099,
      "step": 395
    },
    {
      "epoch": 1.87,
      "learning_rate": 2e-05,
      "loss": 0.3061,
      "step": 396
    },
    {
      "epoch": 1.87,
      "learning_rate": 2e-05,
      "loss": 0.3067,
      "step": 397
    },
    {
      "epoch": 1.88,
      "learning_rate": 2e-05,
      "loss": 0.3099,
      "step": 398
    },
    {
      "epoch": 1.88,
      "learning_rate": 2e-05,
      "loss": 0.3028,
      "step": 399
    },
    {
      "epoch": 1.89,
      "learning_rate": 2e-05,
      "loss": 0.3004,
      "step": 400
    },
    {
      "epoch": 1.89,
      "learning_rate": 2e-05,
      "loss": 0.2966,
      "step": 401
    },
    {
      "epoch": 1.9,
      "learning_rate": 2e-05,
      "loss": 0.2976,
      "step": 402
    },
    {
      "epoch": 1.9,
      "learning_rate": 2e-05,
      "loss": 0.3069,
      "step": 403
    },
    {
      "epoch": 1.91,
      "learning_rate": 2e-05,
      "loss": 0.3069,
      "step": 404
    },
    {
      "epoch": 1.91,
      "learning_rate": 2e-05,
      "loss": 0.3017,
      "step": 405
    },
    {
      "epoch": 1.92,
      "learning_rate": 2e-05,
      "loss": 0.3063,
      "step": 406
    },
    {
      "epoch": 1.92,
      "learning_rate": 2e-05,
      "loss": 0.3023,
      "step": 407
    },
    {
      "epoch": 1.92,
      "learning_rate": 2e-05,
      "loss": 0.3038,
      "step": 408
    },
    {
      "epoch": 1.93,
      "learning_rate": 2e-05,
      "loss": 0.2916,
      "step": 409
    },
    {
      "epoch": 1.93,
      "learning_rate": 2e-05,
      "loss": 0.3027,
      "step": 410
    },
    {
      "epoch": 1.94,
      "learning_rate": 2e-05,
      "loss": 0.3093,
      "step": 411
    },
    {
      "epoch": 1.94,
      "learning_rate": 2e-05,
      "loss": 0.3067,
      "step": 412
    },
    {
      "epoch": 1.95,
      "learning_rate": 2e-05,
      "loss": 0.3059,
      "step": 413
    },
    {
      "epoch": 1.95,
      "learning_rate": 2e-05,
      "loss": 0.3058,
      "step": 414
    },
    {
      "epoch": 1.96,
      "learning_rate": 2e-05,
      "loss": 0.2997,
      "step": 415
    },
    {
      "epoch": 1.96,
      "learning_rate": 2e-05,
      "loss": 0.2883,
      "step": 416
    },
    {
      "epoch": 1.97,
      "learning_rate": 2e-05,
      "loss": 0.2973,
      "step": 417
    },
    {
      "epoch": 1.97,
      "learning_rate": 2e-05,
      "loss": 0.2957,
      "step": 418
    },
    {
      "epoch": 1.98,
      "learning_rate": 2e-05,
      "loss": 0.2999,
      "step": 419
    },
    {
      "epoch": 1.98,
      "learning_rate": 2e-05,
      "loss": 0.2907,
      "step": 420
    },
    {
      "epoch": 1.99,
      "learning_rate": 2e-05,
      "loss": 0.2932,
      "step": 421
    },
    {
      "epoch": 1.99,
      "learning_rate": 2e-05,
      "loss": 0.3027,
      "step": 422
    },
    {
      "epoch": 2.0,
      "learning_rate": 2e-05,
      "loss": 0.2969,
      "step": 423
    },
    {
      "epoch": 2.0,
      "learning_rate": 2e-05,
      "loss": 0.285,
      "step": 424
    },
    {
      "epoch": 2.0,
      "learning_rate": 2e-05,
      "loss": 0.2538,
      "step": 425
    },
    {
      "epoch": 2.01,
      "learning_rate": 2e-05,
      "loss": 0.246,
      "step": 426
    },
    {
      "epoch": 2.01,
      "learning_rate": 2e-05,
      "loss": 0.2431,
      "step": 427
    },
    {
      "epoch": 2.02,
      "learning_rate": 2e-05,
      "loss": 0.2446,
      "step": 428
    },
    {
      "epoch": 2.02,
      "learning_rate": 2e-05,
      "loss": 0.2447,
      "step": 429
    },
    {
      "epoch": 2.03,
      "learning_rate": 2e-05,
      "loss": 0.2449,
      "step": 430
    },
    {
      "epoch": 2.03,
      "learning_rate": 2e-05,
      "loss": 0.2403,
      "step": 431
    },
    {
      "epoch": 2.04,
      "learning_rate": 2e-05,
      "loss": 0.2478,
      "step": 432
    },
    {
      "epoch": 2.04,
      "learning_rate": 2e-05,
      "loss": 0.2383,
      "step": 433
    },
    {
      "epoch": 2.05,
      "learning_rate": 2e-05,
      "loss": 0.2383,
      "step": 434
    },
    {
      "epoch": 2.05,
      "learning_rate": 2e-05,
      "loss": 0.2494,
      "step": 435
    },
    {
      "epoch": 2.06,
      "learning_rate": 2e-05,
      "loss": 0.2345,
      "step": 436
    },
    {
      "epoch": 2.06,
      "learning_rate": 2e-05,
      "loss": 0.2443,
      "step": 437
    },
    {
      "epoch": 2.07,
      "learning_rate": 2e-05,
      "loss": 0.2506,
      "step": 438
    },
    {
      "epoch": 2.07,
      "learning_rate": 2e-05,
      "loss": 0.2478,
      "step": 439
    },
    {
      "epoch": 2.08,
      "learning_rate": 2e-05,
      "loss": 0.2407,
      "step": 440
    },
    {
      "epoch": 2.08,
      "learning_rate": 2e-05,
      "loss": 0.2476,
      "step": 441
    },
    {
      "epoch": 2.08,
      "learning_rate": 2e-05,
      "loss": 0.2448,
      "step": 442
    },
    {
      "epoch": 2.09,
      "learning_rate": 2e-05,
      "loss": 0.25,
      "step": 443
    },
    {
      "epoch": 2.09,
      "learning_rate": 2e-05,
      "loss": 0.2377,
      "step": 444
    },
    {
      "epoch": 2.1,
      "learning_rate": 2e-05,
      "loss": 0.2442,
      "step": 445
    },
    {
      "epoch": 2.1,
      "learning_rate": 2e-05,
      "loss": 0.2458,
      "step": 446
    },
    {
      "epoch": 2.11,
      "learning_rate": 2e-05,
      "loss": 0.2436,
      "step": 447
    },
    {
      "epoch": 2.11,
      "learning_rate": 2e-05,
      "loss": 0.2599,
      "step": 448
    },
    {
      "epoch": 2.12,
      "learning_rate": 2e-05,
      "loss": 0.2447,
      "step": 449
    },
    {
      "epoch": 2.12,
      "learning_rate": 2e-05,
      "loss": 0.2504,
      "step": 450
    },
    {
      "epoch": 2.13,
      "learning_rate": 2e-05,
      "loss": 0.2489,
      "step": 451
    },
    {
      "epoch": 2.13,
      "learning_rate": 2e-05,
      "loss": 0.2413,
      "step": 452
    },
    {
      "epoch": 2.14,
      "learning_rate": 2e-05,
      "loss": 0.2382,
      "step": 453
    },
    {
      "epoch": 2.14,
      "learning_rate": 2e-05,
      "loss": 0.2413,
      "step": 454
    },
    {
      "epoch": 2.15,
      "learning_rate": 2e-05,
      "loss": 0.239,
      "step": 455
    },
    {
      "epoch": 2.15,
      "learning_rate": 2e-05,
      "loss": 0.2441,
      "step": 456
    },
    {
      "epoch": 2.16,
      "learning_rate": 2e-05,
      "loss": 0.2362,
      "step": 457
    },
    {
      "epoch": 2.16,
      "learning_rate": 2e-05,
      "loss": 0.2527,
      "step": 458
    },
    {
      "epoch": 2.17,
      "learning_rate": 2e-05,
      "loss": 0.2387,
      "step": 459
    },
    {
      "epoch": 2.17,
      "learning_rate": 2e-05,
      "loss": 0.2449,
      "step": 460
    },
    {
      "epoch": 2.17,
      "learning_rate": 2e-05,
      "loss": 0.2439,
      "step": 461
    },
    {
      "epoch": 2.18,
      "learning_rate": 2e-05,
      "loss": 0.2386,
      "step": 462
    },
    {
      "epoch": 2.18,
      "learning_rate": 2e-05,
      "loss": 0.2386,
      "step": 463
    },
    {
      "epoch": 2.19,
      "learning_rate": 2e-05,
      "loss": 0.2421,
      "step": 464
    },
    {
      "epoch": 2.19,
      "learning_rate": 2e-05,
      "loss": 0.252,
      "step": 465
    },
    {
      "epoch": 2.2,
      "learning_rate": 2e-05,
      "loss": 0.241,
      "step": 466
    },
    {
      "epoch": 2.2,
      "learning_rate": 2e-05,
      "loss": 0.242,
      "step": 467
    },
    {
      "epoch": 2.21,
      "learning_rate": 2e-05,
      "loss": 0.2454,
      "step": 468
    },
    {
      "epoch": 2.21,
      "learning_rate": 2e-05,
      "loss": 0.2456,
      "step": 469
    },
    {
      "epoch": 2.22,
      "learning_rate": 2e-05,
      "loss": 0.2539,
      "step": 470
    },
    {
      "epoch": 2.22,
      "learning_rate": 2e-05,
      "loss": 0.2415,
      "step": 471
    },
    {
      "epoch": 2.23,
      "learning_rate": 2e-05,
      "loss": 0.249,
      "step": 472
    },
    {
      "epoch": 2.23,
      "learning_rate": 2e-05,
      "loss": 0.2524,
      "step": 473
    },
    {
      "epoch": 2.24,
      "learning_rate": 2e-05,
      "loss": 0.2551,
      "step": 474
    },
    {
      "epoch": 2.24,
      "learning_rate": 2e-05,
      "loss": 0.2417,
      "step": 475
    },
    {
      "epoch": 2.25,
      "learning_rate": 2e-05,
      "loss": 0.2464,
      "step": 476
    },
    {
      "epoch": 2.25,
      "learning_rate": 2e-05,
      "loss": 0.246,
      "step": 477
    },
    {
      "epoch": 2.25,
      "learning_rate": 2e-05,
      "loss": 0.2473,
      "step": 478
    },
    {
      "epoch": 2.26,
      "learning_rate": 2e-05,
      "loss": 0.2535,
      "step": 479
    },
    {
      "epoch": 2.26,
      "learning_rate": 2e-05,
      "loss": 0.2525,
      "step": 480
    },
    {
      "epoch": 2.27,
      "learning_rate": 2e-05,
      "loss": 0.2479,
      "step": 481
    },
    {
      "epoch": 2.27,
      "learning_rate": 2e-05,
      "loss": 0.2486,
      "step": 482
    },
    {
      "epoch": 2.28,
      "learning_rate": 2e-05,
      "loss": 0.2427,
      "step": 483
    },
    {
      "epoch": 2.28,
      "learning_rate": 2e-05,
      "loss": 0.2528,
      "step": 484
    },
    {
      "epoch": 2.29,
      "learning_rate": 2e-05,
      "loss": 0.2447,
      "step": 485
    },
    {
      "epoch": 2.29,
      "learning_rate": 2e-05,
      "loss": 0.2437,
      "step": 486
    },
    {
      "epoch": 2.3,
      "learning_rate": 2e-05,
      "loss": 0.2491,
      "step": 487
    },
    {
      "epoch": 2.3,
      "learning_rate": 2e-05,
      "loss": 0.2492,
      "step": 488
    },
    {
      "epoch": 2.31,
      "learning_rate": 2e-05,
      "loss": 0.2417,
      "step": 489
    },
    {
      "epoch": 2.31,
      "learning_rate": 2e-05,
      "loss": 0.2436,
      "step": 490
    },
    {
      "epoch": 2.32,
      "learning_rate": 2e-05,
      "loss": 0.2589,
      "step": 491
    },
    {
      "epoch": 2.32,
      "learning_rate": 2e-05,
      "loss": 0.2441,
      "step": 492
    },
    {
      "epoch": 2.33,
      "learning_rate": 2e-05,
      "loss": 0.2406,
      "step": 493
    },
    {
      "epoch": 2.33,
      "learning_rate": 2e-05,
      "loss": 0.2464,
      "step": 494
    },
    {
      "epoch": 2.33,
      "learning_rate": 2e-05,
      "loss": 0.2375,
      "step": 495
    },
    {
      "epoch": 2.34,
      "learning_rate": 2e-05,
      "loss": 0.2449,
      "step": 496
    },
    {
      "epoch": 2.34,
      "learning_rate": 2e-05,
      "loss": 0.2458,
      "step": 497
    },
    {
      "epoch": 2.35,
      "learning_rate": 2e-05,
      "loss": 0.2386,
      "step": 498
    },
    {
      "epoch": 2.35,
      "learning_rate": 2e-05,
      "loss": 0.2471,
      "step": 499
    },
    {
      "epoch": 2.36,
      "learning_rate": 2e-05,
      "loss": 0.2401,
      "step": 500
    },
    {
      "epoch": 2.36,
      "learning_rate": 2e-05,
      "loss": 0.2435,
      "step": 501
    },
    {
      "epoch": 2.37,
      "learning_rate": 2e-05,
      "loss": 0.2436,
      "step": 502
    },
    {
      "epoch": 2.37,
      "learning_rate": 2e-05,
      "loss": 0.2401,
      "step": 503
    },
    {
      "epoch": 2.38,
      "learning_rate": 2e-05,
      "loss": 0.2541,
      "step": 504
    },
    {
      "epoch": 2.38,
      "learning_rate": 2e-05,
      "loss": 0.2466,
      "step": 505
    },
    {
      "epoch": 2.39,
      "learning_rate": 2e-05,
      "loss": 0.2501,
      "step": 506
    },
    {
      "epoch": 2.39,
      "learning_rate": 2e-05,
      "loss": 0.2431,
      "step": 507
    },
    {
      "epoch": 2.4,
      "learning_rate": 2e-05,
      "loss": 0.2452,
      "step": 508
    },
    {
      "epoch": 2.4,
      "learning_rate": 2e-05,
      "loss": 0.2413,
      "step": 509
    },
    {
      "epoch": 2.41,
      "learning_rate": 2e-05,
      "loss": 0.2449,
      "step": 510
    },
    {
      "epoch": 2.41,
      "learning_rate": 2e-05,
      "loss": 0.2384,
      "step": 511
    },
    {
      "epoch": 2.42,
      "learning_rate": 2e-05,
      "loss": 0.2474,
      "step": 512
    },
    {
      "epoch": 2.42,
      "learning_rate": 2e-05,
      "loss": 0.2412,
      "step": 513
    },
    {
      "epoch": 2.42,
      "learning_rate": 2e-05,
      "loss": 0.2437,
      "step": 514
    },
    {
      "epoch": 2.43,
      "learning_rate": 2e-05,
      "loss": 0.2482,
      "step": 515
    },
    {
      "epoch": 2.43,
      "learning_rate": 2e-05,
      "loss": 0.2431,
      "step": 516
    },
    {
      "epoch": 2.44,
      "learning_rate": 2e-05,
      "loss": 0.2472,
      "step": 517
    },
    {
      "epoch": 2.44,
      "learning_rate": 2e-05,
      "loss": 0.2431,
      "step": 518
    },
    {
      "epoch": 2.45,
      "learning_rate": 2e-05,
      "loss": 0.2422,
      "step": 519
    },
    {
      "epoch": 2.45,
      "learning_rate": 2e-05,
      "loss": 0.247,
      "step": 520
    },
    {
      "epoch": 2.46,
      "learning_rate": 2e-05,
      "loss": 0.2455,
      "step": 521
    },
    {
      "epoch": 2.46,
      "learning_rate": 2e-05,
      "loss": 0.2448,
      "step": 522
    },
    {
      "epoch": 2.47,
      "learning_rate": 2e-05,
      "loss": 0.2428,
      "step": 523
    },
    {
      "epoch": 2.47,
      "learning_rate": 2e-05,
      "loss": 0.243,
      "step": 524
    },
    {
      "epoch": 2.48,
      "learning_rate": 2e-05,
      "loss": 0.235,
      "step": 525
    },
    {
      "epoch": 2.48,
      "learning_rate": 2e-05,
      "loss": 0.2415,
      "step": 526
    },
    {
      "epoch": 2.49,
      "learning_rate": 2e-05,
      "loss": 0.2447,
      "step": 527
    },
    {
      "epoch": 2.49,
      "learning_rate": 2e-05,
      "loss": 0.2443,
      "step": 528
    },
    {
      "epoch": 2.5,
      "learning_rate": 2e-05,
      "loss": 0.2391,
      "step": 529
    },
    {
      "epoch": 2.5,
      "learning_rate": 2e-05,
      "loss": 0.2398,
      "step": 530
    },
    {
      "epoch": 2.5,
      "learning_rate": 2e-05,
      "loss": 0.244,
      "step": 531
    },
    {
      "epoch": 2.51,
      "learning_rate": 2e-05,
      "loss": 0.2404,
      "step": 532
    },
    {
      "epoch": 2.51,
      "learning_rate": 2e-05,
      "loss": 0.2416,
      "step": 533
    },
    {
      "epoch": 2.52,
      "learning_rate": 2e-05,
      "loss": 0.2401,
      "step": 534
    },
    {
      "epoch": 2.52,
      "learning_rate": 2e-05,
      "loss": 0.2449,
      "step": 535
    },
    {
      "epoch": 2.53,
      "learning_rate": 2e-05,
      "loss": 0.2407,
      "step": 536
    },
    {
      "epoch": 2.53,
      "learning_rate": 2e-05,
      "loss": 0.2411,
      "step": 537
    },
    {
      "epoch": 2.54,
      "learning_rate": 2e-05,
      "loss": 0.2376,
      "step": 538
    },
    {
      "epoch": 2.54,
      "learning_rate": 2e-05,
      "loss": 0.2444,
      "step": 539
    },
    {
      "epoch": 2.55,
      "learning_rate": 2e-05,
      "loss": 0.2436,
      "step": 540
    },
    {
      "epoch": 2.55,
      "learning_rate": 2e-05,
      "loss": 0.2488,
      "step": 541
    },
    {
      "epoch": 2.56,
      "learning_rate": 2e-05,
      "loss": 0.2377,
      "step": 542
    },
    {
      "epoch": 2.56,
      "learning_rate": 2e-05,
      "loss": 0.2404,
      "step": 543
    },
    {
      "epoch": 2.57,
      "learning_rate": 2e-05,
      "loss": 0.2385,
      "step": 544
    },
    {
      "epoch": 2.57,
      "learning_rate": 2e-05,
      "loss": 0.2401,
      "step": 545
    },
    {
      "epoch": 2.58,
      "learning_rate": 2e-05,
      "loss": 0.2363,
      "step": 546
    },
    {
      "epoch": 2.58,
      "learning_rate": 2e-05,
      "loss": 0.2403,
      "step": 547
    },
    {
      "epoch": 2.58,
      "learning_rate": 2e-05,
      "loss": 0.2344,
      "step": 548
    },
    {
      "epoch": 2.59,
      "learning_rate": 2e-05,
      "loss": 0.2481,
      "step": 549
    },
    {
      "epoch": 2.59,
      "learning_rate": 2e-05,
      "loss": 0.2392,
      "step": 550
    },
    {
      "epoch": 2.6,
      "learning_rate": 2e-05,
      "loss": 0.2475,
      "step": 551
    },
    {
      "epoch": 2.6,
      "learning_rate": 2e-05,
      "loss": 0.2436,
      "step": 552
    },
    {
      "epoch": 2.61,
      "learning_rate": 2e-05,
      "loss": 0.242,
      "step": 553
    },
    {
      "epoch": 2.61,
      "learning_rate": 2e-05,
      "loss": 0.2427,
      "step": 554
    },
    {
      "epoch": 2.62,
      "learning_rate": 2e-05,
      "loss": 0.2316,
      "step": 555
    },
    {
      "epoch": 2.62,
      "learning_rate": 2e-05,
      "loss": 0.2326,
      "step": 556
    },
    {
      "epoch": 2.63,
      "learning_rate": 2e-05,
      "loss": 0.2406,
      "step": 557
    },
    {
      "epoch": 2.63,
      "learning_rate": 2e-05,
      "loss": 0.2418,
      "step": 558
    },
    {
      "epoch": 2.64,
      "learning_rate": 2e-05,
      "loss": 0.2402,
      "step": 559
    },
    {
      "epoch": 2.64,
      "learning_rate": 2e-05,
      "loss": 0.243,
      "step": 560
    },
    {
      "epoch": 2.65,
      "learning_rate": 2e-05,
      "loss": 0.2365,
      "step": 561
    },
    {
      "epoch": 2.65,
      "learning_rate": 2e-05,
      "loss": 0.241,
      "step": 562
    },
    {
      "epoch": 2.66,
      "learning_rate": 2e-05,
      "loss": 0.245,
      "step": 563
    },
    {
      "epoch": 2.66,
      "learning_rate": 2e-05,
      "loss": 0.2303,
      "step": 564
    },
    {
      "epoch": 2.67,
      "learning_rate": 2e-05,
      "loss": 0.2415,
      "step": 565
    },
    {
      "epoch": 2.67,
      "learning_rate": 2e-05,
      "loss": 0.234,
      "step": 566
    },
    {
      "epoch": 2.67,
      "learning_rate": 2e-05,
      "loss": 0.2478,
      "step": 567
    },
    {
      "epoch": 2.68,
      "learning_rate": 2e-05,
      "loss": 0.2538,
      "step": 568
    },
    {
      "epoch": 2.68,
      "learning_rate": 2e-05,
      "loss": 0.2353,
      "step": 569
    },
    {
      "epoch": 2.69,
      "learning_rate": 2e-05,
      "loss": 0.2452,
      "step": 570
    },
    {
      "epoch": 2.69,
      "learning_rate": 2e-05,
      "loss": 0.2369,
      "step": 571
    },
    {
      "epoch": 2.7,
      "learning_rate": 2e-05,
      "loss": 0.2386,
      "step": 572
    },
    {
      "epoch": 2.7,
      "learning_rate": 2e-05,
      "loss": 0.2361,
      "step": 573
    },
    {
      "epoch": 2.71,
      "learning_rate": 2e-05,
      "loss": 0.2329,
      "step": 574
    },
    {
      "epoch": 2.71,
      "learning_rate": 2e-05,
      "loss": 0.2316,
      "step": 575
    },
    {
      "epoch": 2.72,
      "learning_rate": 2e-05,
      "loss": 0.2354,
      "step": 576
    },
    {
      "epoch": 2.72,
      "learning_rate": 2e-05,
      "loss": 0.2396,
      "step": 577
    },
    {
      "epoch": 2.73,
      "learning_rate": 2e-05,
      "loss": 0.2318,
      "step": 578
    },
    {
      "epoch": 2.73,
      "learning_rate": 2e-05,
      "loss": 0.2262,
      "step": 579
    },
    {
      "epoch": 2.74,
      "learning_rate": 2e-05,
      "loss": 0.2379,
      "step": 580
    },
    {
      "epoch": 2.74,
      "learning_rate": 2e-05,
      "loss": 0.2392,
      "step": 581
    },
    {
      "epoch": 2.75,
      "learning_rate": 2e-05,
      "loss": 0.2328,
      "step": 582
    },
    {
      "epoch": 2.75,
      "learning_rate": 2e-05,
      "loss": 0.2336,
      "step": 583
    },
    {
      "epoch": 2.75,
      "learning_rate": 2e-05,
      "loss": 0.2369,
      "step": 584
    },
    {
      "epoch": 2.76,
      "learning_rate": 2e-05,
      "loss": 0.2349,
      "step": 585
    },
    {
      "epoch": 2.76,
      "learning_rate": 2e-05,
      "loss": 0.2344,
      "step": 586
    },
    {
      "epoch": 2.77,
      "learning_rate": 2e-05,
      "loss": 0.2384,
      "step": 587
    },
    {
      "epoch": 2.77,
      "learning_rate": 2e-05,
      "loss": 0.2434,
      "step": 588
    },
    {
      "epoch": 2.78,
      "learning_rate": 2e-05,
      "loss": 0.2437,
      "step": 589
    },
    {
      "epoch": 2.78,
      "learning_rate": 2e-05,
      "loss": 0.2354,
      "step": 590
    },
    {
      "epoch": 2.79,
      "learning_rate": 2e-05,
      "loss": 0.2279,
      "step": 591
    },
    {
      "epoch": 2.79,
      "learning_rate": 2e-05,
      "loss": 0.2378,
      "step": 592
    },
    {
      "epoch": 2.8,
      "learning_rate": 2e-05,
      "loss": 0.2338,
      "step": 593
    },
    {
      "epoch": 2.8,
      "learning_rate": 2e-05,
      "loss": 0.2279,
      "step": 594
    },
    {
      "epoch": 2.81,
      "learning_rate": 2e-05,
      "loss": 0.2332,
      "step": 595
    },
    {
      "epoch": 2.81,
      "learning_rate": 2e-05,
      "loss": 0.2313,
      "step": 596
    },
    {
      "epoch": 2.82,
      "learning_rate": 2e-05,
      "loss": 0.2455,
      "step": 597
    },
    {
      "epoch": 2.82,
      "learning_rate": 2e-05,
      "loss": 0.2375,
      "step": 598
    },
    {
      "epoch": 2.83,
      "learning_rate": 2e-05,
      "loss": 0.2313,
      "step": 599
    },
    {
      "epoch": 2.83,
      "learning_rate": 2e-05,
      "loss": 0.2379,
      "step": 600
    },
    {
      "epoch": 2.83,
      "learning_rate": 2e-05,
      "loss": 0.2303,
      "step": 601
    },
    {
      "epoch": 2.84,
      "learning_rate": 2e-05,
      "loss": 0.235,
      "step": 602
    },
    {
      "epoch": 2.84,
      "learning_rate": 2e-05,
      "loss": 0.2288,
      "step": 603
    },
    {
      "epoch": 2.85,
      "learning_rate": 2e-05,
      "loss": 0.2257,
      "step": 604
    },
    {
      "epoch": 2.85,
      "learning_rate": 2e-05,
      "loss": 0.2302,
      "step": 605
    },
    {
      "epoch": 2.86,
      "learning_rate": 2e-05,
      "loss": 0.2326,
      "step": 606
    },
    {
      "epoch": 2.86,
      "learning_rate": 2e-05,
      "loss": 0.2324,
      "step": 607
    },
    {
      "epoch": 2.87,
      "learning_rate": 2e-05,
      "loss": 0.2352,
      "step": 608
    },
    {
      "epoch": 2.87,
      "learning_rate": 2e-05,
      "loss": 0.2319,
      "step": 609
    },
    {
      "epoch": 2.88,
      "learning_rate": 2e-05,
      "loss": 0.2357,
      "step": 610
    },
    {
      "epoch": 2.88,
      "learning_rate": 2e-05,
      "loss": 0.2295,
      "step": 611
    },
    {
      "epoch": 2.89,
      "learning_rate": 2e-05,
      "loss": 0.2298,
      "step": 612
    },
    {
      "epoch": 2.89,
      "learning_rate": 2e-05,
      "loss": 0.2216,
      "step": 613
    },
    {
      "epoch": 2.9,
      "learning_rate": 2e-05,
      "loss": 0.2216,
      "step": 614
    },
    {
      "epoch": 2.9,
      "learning_rate": 2e-05,
      "loss": 0.2271,
      "step": 615
    },
    {
      "epoch": 2.91,
      "learning_rate": 2e-05,
      "loss": 0.2256,
      "step": 616
    },
    {
      "epoch": 2.91,
      "learning_rate": 2e-05,
      "loss": 0.2331,
      "step": 617
    },
    {
      "epoch": 2.92,
      "learning_rate": 2e-05,
      "loss": 0.2338,
      "step": 618
    },
    {
      "epoch": 2.92,
      "learning_rate": 2e-05,
      "loss": 0.2256,
      "step": 619
    },
    {
      "epoch": 2.92,
      "learning_rate": 2e-05,
      "loss": 0.2283,
      "step": 620
    },
    {
      "epoch": 2.93,
      "learning_rate": 2e-05,
      "loss": 0.2316,
      "step": 621
    },
    {
      "epoch": 2.93,
      "learning_rate": 2e-05,
      "loss": 0.22,
      "step": 622
    },
    {
      "epoch": 2.94,
      "learning_rate": 2e-05,
      "loss": 0.2232,
      "step": 623
    },
    {
      "epoch": 2.94,
      "learning_rate": 2e-05,
      "loss": 0.2262,
      "step": 624
    },
    {
      "epoch": 2.95,
      "learning_rate": 2e-05,
      "loss": 0.2274,
      "step": 625
    },
    {
      "epoch": 2.95,
      "learning_rate": 2e-05,
      "loss": 0.2333,
      "step": 626
    },
    {
      "epoch": 2.96,
      "learning_rate": 2e-05,
      "loss": 0.2292,
      "step": 627
    },
    {
      "epoch": 2.96,
      "learning_rate": 2e-05,
      "loss": 0.2308,
      "step": 628
    },
    {
      "epoch": 2.97,
      "learning_rate": 2e-05,
      "loss": 0.2293,
      "step": 629
    },
    {
      "epoch": 2.97,
      "learning_rate": 2e-05,
      "loss": 0.2248,
      "step": 630
    },
    {
      "epoch": 2.98,
      "learning_rate": 2e-05,
      "loss": 0.2336,
      "step": 631
    },
    {
      "epoch": 2.98,
      "learning_rate": 2e-05,
      "loss": 0.2292,
      "step": 632
    },
    {
      "epoch": 2.99,
      "learning_rate": 2e-05,
      "loss": 0.2293,
      "step": 633
    },
    {
      "epoch": 2.99,
      "learning_rate": 2e-05,
      "loss": 0.2204,
      "step": 634
    },
    {
      "epoch": 3.0,
      "learning_rate": 2e-05,
      "loss": 0.221,
      "step": 635
    },
    {
      "epoch": 3.0,
      "learning_rate": 2e-05,
      "loss": 0.2101,
      "step": 636
    },
    {
      "epoch": 3.0,
      "step": 636,
      "total_flos": 1068955157921792.0,
      "train_loss": 0.3158094177268586,
      "train_runtime": 14494.9673,
      "train_samples_per_second": 5.603,
      "train_steps_per_second": 0.044
    }
  ],
  "max_steps": 636,
  "num_train_epochs": 3,
  "total_flos": 1068955157921792.0,
  "trial_name": null,
  "trial_params": null
}