{
  "best_metric": 0.9114114114114115,
  "best_model_checkpoint": "test_trainer/checkpoint-3000",
  "epoch": 6.9614168247944335,
  "global_step": 5500,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "learning_rate": 4.999095840867993e-05,
      "loss": 0.6609,
      "step": 1
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.998191681735986e-05,
      "loss": 0.6074,
      "step": 2
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9972875226039784e-05,
      "loss": 0.6241,
      "step": 3
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9963833634719715e-05,
      "loss": 0.5636,
      "step": 4
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.995479204339964e-05,
      "loss": 0.4803,
      "step": 5
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.994575045207957e-05,
      "loss": 0.51,
      "step": 6
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.99367088607595e-05,
      "loss": 0.5337,
      "step": 7
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.992766726943942e-05,
      "loss": 0.4156,
      "step": 8
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.991862567811935e-05,
      "loss": 0.4586,
      "step": 9
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.990958408679928e-05,
      "loss": 0.5977,
      "step": 10
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.990054249547921e-05,
      "loss": 0.3624,
      "step": 11
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9891500904159134e-05,
      "loss": 0.3493,
      "step": 12
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9882459312839066e-05,
      "loss": 0.3395,
      "step": 13
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.987341772151899e-05,
      "loss": 0.4243,
      "step": 14
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.986437613019892e-05,
      "loss": 0.4844,
      "step": 15
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.985533453887885e-05,
      "loss": 0.268,
      "step": 16
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.984629294755877e-05,
      "loss": 0.3212,
      "step": 17
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9837251356238704e-05,
      "loss": 0.4282,
      "step": 18
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.982820976491863e-05,
      "loss": 0.4145,
      "step": 19
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.981916817359856e-05,
      "loss": 0.4906,
      "step": 20
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9810126582278485e-05,
      "loss": 0.492,
      "step": 21
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.980108499095841e-05,
      "loss": 0.5006,
      "step": 22
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.979204339963834e-05,
      "loss": 0.317,
      "step": 23
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9783001808318267e-05,
      "loss": 0.487,
      "step": 24
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.977396021699819e-05,
      "loss": 0.2456,
      "step": 25
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.976491862567812e-05,
      "loss": 0.4812,
      "step": 26
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.975587703435805e-05,
      "loss": 0.4794,
      "step": 27
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.974683544303797e-05,
      "loss": 0.2499,
      "step": 28
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9737793851717904e-05,
      "loss": 0.4182,
      "step": 29
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.972875226039783e-05,
      "loss": 0.41,
      "step": 30
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.971971066907776e-05,
      "loss": 0.3072,
      "step": 31
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9710669077757686e-05,
      "loss": 0.2699,
      "step": 32
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.970162748643761e-05,
      "loss": 0.404,
      "step": 33
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.969258589511754e-05,
      "loss": 0.3723,
      "step": 34
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.968354430379747e-05,
      "loss": 0.4503,
      "step": 35
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.96745027124774e-05,
      "loss": 0.4717,
      "step": 36
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9665461121157323e-05,
      "loss": 0.1534,
      "step": 37
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9656419529837255e-05,
      "loss": 0.5552,
      "step": 38
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.964737793851718e-05,
      "loss": 0.4708,
      "step": 39
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9638336347197105e-05,
      "loss": 0.6868,
      "step": 40
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9629294755877036e-05,
      "loss": 0.5114,
      "step": 41
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.962025316455696e-05,
      "loss": 0.4045,
      "step": 42
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.961121157323689e-05,
      "loss": 0.3987,
      "step": 43
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.960216998191682e-05,
      "loss": 0.4725,
      "step": 44
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.959312839059675e-05,
      "loss": 0.3961,
      "step": 45
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9584086799276674e-05,
      "loss": 0.3516,
      "step": 46
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9575045207956606e-05,
      "loss": 0.3351,
      "step": 47
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.956600361663653e-05,
      "loss": 0.359,
      "step": 48
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9556962025316456e-05,
      "loss": 0.4319,
      "step": 49
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.954792043399639e-05,
      "loss": 0.333,
      "step": 50
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.953887884267631e-05,
      "loss": 0.5152,
      "step": 51
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9529837251356244e-05,
      "loss": 0.3084,
      "step": 52
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.952079566003617e-05,
      "loss": 0.4306,
      "step": 53
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.95117540687161e-05,
      "loss": 0.2049,
      "step": 54
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9502712477396025e-05,
      "loss": 0.2995,
      "step": 55
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.949367088607595e-05,
      "loss": 0.3411,
      "step": 56
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.948462929475588e-05,
      "loss": 0.3542,
      "step": 57
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9475587703435806e-05,
      "loss": 0.4064,
      "step": 58
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.946654611211574e-05,
      "loss": 0.4607,
      "step": 59
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.945750452079566e-05,
      "loss": 0.3895,
      "step": 60
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9448462929475594e-05,
      "loss": 0.3048,
      "step": 61
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.943942133815552e-05,
      "loss": 0.3737,
      "step": 62
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.943037974683545e-05,
      "loss": 0.2722,
      "step": 63
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9421338155515376e-05,
      "loss": 0.4133,
      "step": 64
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.94122965641953e-05,
      "loss": 0.4533,
      "step": 65
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.940325497287523e-05,
      "loss": 0.3968,
      "step": 66
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.939421338155516e-05,
      "loss": 0.4046,
      "step": 67
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.938517179023508e-05,
      "loss": 0.3709,
      "step": 68
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9376130198915014e-05,
      "loss": 0.4248,
      "step": 69
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.936708860759494e-05,
      "loss": 0.4011,
      "step": 70
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.935804701627486e-05,
      "loss": 0.3844,
      "step": 71
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9349005424954795e-05,
      "loss": 0.2778,
      "step": 72
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.933996383363472e-05,
      "loss": 0.4103,
      "step": 73
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9330922242314645e-05,
      "loss": 0.4219,
      "step": 74
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9321880650994576e-05,
      "loss": 0.2543,
      "step": 75
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.93128390596745e-05,
      "loss": 0.4822,
      "step": 76
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.930379746835443e-05,
      "loss": 0.4056,
      "step": 77
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.929475587703436e-05,
      "loss": 0.4227,
      "step": 78
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.928571428571429e-05,
      "loss": 0.3085,
      "step": 79
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.9276672694394214e-05,
      "loss": 0.4621,
      "step": 80
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.926763110307414e-05,
      "loss": 0.4106,
      "step": 81
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.925858951175407e-05,
      "loss": 0.2682,
      "step": 82
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.9249547920433995e-05,
      "loss": 0.3003,
      "step": 83
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.924050632911393e-05,
      "loss": 0.2767,
      "step": 84
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.923146473779385e-05,
      "loss": 0.2611,
      "step": 85
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.9222423146473783e-05,
      "loss": 0.3287,
      "step": 86
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.921338155515371e-05,
      "loss": 0.5682,
      "step": 87
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.920433996383363e-05,
      "loss": 0.382,
      "step": 88
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.9195298372513565e-05,
      "loss": 0.5605,
      "step": 89
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.918625678119349e-05,
      "loss": 0.4982,
      "step": 90
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.917721518987342e-05,
      "loss": 0.3121,
      "step": 91
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.9168173598553346e-05,
      "loss": 0.508,
      "step": 92
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.915913200723328e-05,
      "loss": 0.2933,
      "step": 93
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.91500904159132e-05,
      "loss": 0.5757,
      "step": 94
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.9141048824593134e-05,
      "loss": 0.4157,
      "step": 95
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.913200723327306e-05,
      "loss": 0.3097,
      "step": 96
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.9122965641952984e-05,
      "loss": 0.3642,
      "step": 97
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.9113924050632915e-05,
      "loss": 0.4281,
      "step": 98
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.910488245931284e-05,
      "loss": 0.3614,
      "step": 99
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.909584086799277e-05,
      "loss": 0.4539,
      "step": 100
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.90867992766727e-05,
      "loss": 0.4231,
      "step": 101
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.907775768535263e-05,
      "loss": 0.2322,
      "step": 102
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.906871609403255e-05,
      "loss": 0.2751,
      "step": 103
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.905967450271248e-05,
      "loss": 0.2266,
      "step": 104
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.905063291139241e-05,
      "loss": 0.3544,
      "step": 105
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.9041591320072335e-05,
      "loss": 0.2705,
      "step": 106
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.9032549728752266e-05,
      "loss": 0.3982,
      "step": 107
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.902350813743219e-05,
      "loss": 0.3368,
      "step": 108
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.901446654611212e-05,
      "loss": 0.2562,
      "step": 109
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.900542495479205e-05,
      "loss": 0.2765,
      "step": 110
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.899638336347198e-05,
      "loss": 0.2969,
      "step": 111
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.8987341772151904e-05,
      "loss": 0.216,
      "step": 112
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.897830018083183e-05,
      "loss": 0.3154,
      "step": 113
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.896925858951176e-05,
      "loss": 0.25,
      "step": 114
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.8960216998191685e-05,
      "loss": 0.2832,
      "step": 115
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.895117540687161e-05,
      "loss": 0.4115,
      "step": 116
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.894213381555154e-05,
      "loss": 0.1538,
      "step": 117
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.893309222423147e-05,
      "loss": 0.2062,
      "step": 118
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.892405063291139e-05,
      "loss": 0.3296,
      "step": 119
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.891500904159132e-05,
      "loss": 0.3381,
      "step": 120
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.890596745027125e-05,
      "loss": 0.3242,
      "step": 121
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.889692585895117e-05,
      "loss": 0.2986,
      "step": 122
    },
    {
      "epoch": 0.16,
      "learning_rate": 4.8887884267631105e-05,
      "loss": 0.2781,
      "step": 123
    },
    {
      "epoch": 0.16,
      "learning_rate": 4.887884267631103e-05,
      "loss": 0.3229,
      "step": 124
    },
    {
      "epoch": 0.16,
      "learning_rate": 4.886980108499096e-05,
      "loss": 0.3422,
      "step": 125
    },
    {
      "epoch": 0.16,
      "learning_rate": 4.8860759493670886e-05,
      "loss": 0.4468,
      "step": 126
    },
    {
      "epoch": 0.16,
      "learning_rate": 4.885171790235082e-05,
      "loss": 0.1982,
      "step": 127
    },
    {
      "epoch": 0.16,
      "learning_rate": 4.884267631103074e-05,
      "loss": 0.3768,
      "step": 128
    },
    {
      "epoch": 0.16,
      "learning_rate": 4.883363471971067e-05,
      "loss": 0.4463,
      "step": 129
    },
    {
      "epoch": 0.16,
      "learning_rate": 4.88245931283906e-05,
      "loss": 0.2539,
      "step": 130
    },
    {
      "epoch": 0.17,
      "learning_rate": 4.8815551537070524e-05,
      "loss": 0.3429,
      "step": 131
    },
    {
      "epoch": 0.17,
      "learning_rate": 4.8806509945750455e-05,
      "loss": 0.3187,
      "step": 132
    },
    {
      "epoch": 0.17,
      "learning_rate": 4.879746835443038e-05,
      "loss": 0.2126,
      "step": 133
    },
    {
      "epoch": 0.17,
      "learning_rate": 4.878842676311031e-05,
      "loss": 0.5705,
      "step": 134
    },
    {
      "epoch": 0.17,
      "learning_rate": 4.8779385171790237e-05,
      "loss": 0.2123,
      "step": 135
    },
    {
      "epoch": 0.17,
      "learning_rate": 4.877034358047016e-05,
      "loss": 0.4429,
      "step": 136
    },
    {
      "epoch": 0.17,
      "learning_rate": 4.876130198915009e-05,
      "loss": 0.4235,
      "step": 137
    },
    {
      "epoch": 0.17,
      "learning_rate": 4.875226039783002e-05,
      "loss": 0.455,
      "step": 138
    },
    {
      "epoch": 0.18,
      "learning_rate": 4.874321880650995e-05,
      "loss": 0.3014,
      "step": 139
    },
    {
      "epoch": 0.18,
      "learning_rate": 4.8734177215189874e-05,
      "loss": 0.4378,
      "step": 140
    },
    {
      "epoch": 0.18,
      "learning_rate": 4.8725135623869806e-05,
      "loss": 0.3749,
      "step": 141
    },
    {
      "epoch": 0.18,
      "learning_rate": 4.871609403254973e-05,
      "loss": 0.2506,
      "step": 142
    },
    {
      "epoch": 0.18,
      "learning_rate": 4.870705244122966e-05,
      "loss": 0.266,
      "step": 143
    },
    {
      "epoch": 0.18,
      "learning_rate": 4.869801084990959e-05,
      "loss": 0.3234,
      "step": 144
    },
    {
      "epoch": 0.18,
      "learning_rate": 4.868896925858951e-05,
      "loss": 0.4741,
      "step": 145
    },
    {
      "epoch": 0.18,
      "learning_rate": 4.8679927667269444e-05,
      "loss": 0.4264,
      "step": 146
    },
    {
      "epoch": 0.19,
      "learning_rate": 4.867088607594937e-05,
      "loss": 0.3886,
      "step": 147
    },
    {
      "epoch": 0.19,
      "learning_rate": 4.86618444846293e-05,
      "loss": 0.3543,
      "step": 148
    },
    {
      "epoch": 0.19,
      "learning_rate": 4.8652802893309225e-05,
      "loss": 0.556,
      "step": 149
    },
    {
      "epoch": 0.19,
      "learning_rate": 4.864376130198916e-05,
      "loss": 0.2091,
      "step": 150
    },
    {
      "epoch": 0.19,
      "learning_rate": 4.863471971066908e-05,
      "loss": 0.3252,
      "step": 151
    },
    {
      "epoch": 0.19,
      "learning_rate": 4.8625678119349006e-05,
      "loss": 0.4047,
      "step": 152
    },
    {
      "epoch": 0.19,
      "learning_rate": 4.861663652802894e-05,
      "loss": 0.2824,
      "step": 153
    },
    {
      "epoch": 0.19,
      "learning_rate": 4.860759493670886e-05,
      "loss": 0.292,
      "step": 154
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.8598553345388795e-05,
      "loss": 0.2083,
      "step": 155
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.858951175406872e-05,
      "loss": 0.4722,
      "step": 156
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.858047016274865e-05,
      "loss": 0.2637,
      "step": 157
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.8571428571428576e-05,
      "loss": 0.2403,
      "step": 158
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.85623869801085e-05,
      "loss": 0.3644,
      "step": 159
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.855334538878843e-05,
      "loss": 0.1988,
      "step": 160
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.854430379746836e-05,
      "loss": 0.4802,
      "step": 161
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.853526220614828e-05,
      "loss": 0.3006,
      "step": 162
    },
    {
      "epoch": 0.21,
      "learning_rate": 4.8526220614828214e-05,
      "loss": 0.2543,
      "step": 163
    },
    {
      "epoch": 0.21,
      "learning_rate": 4.851717902350814e-05,
      "loss": 0.3161,
      "step": 164
    },
    {
      "epoch": 0.21,
      "learning_rate": 4.850813743218806e-05,
      "loss": 0.2277,
      "step": 165
    },
    {
      "epoch": 0.21,
      "learning_rate": 4.8499095840867995e-05,
      "loss": 0.1347,
      "step": 166
    },
    {
      "epoch": 0.21,
      "learning_rate": 4.849005424954792e-05,
      "loss": 0.2452,
      "step": 167
    },
    {
      "epoch": 0.21,
      "learning_rate": 4.8481012658227845e-05,
      "loss": 0.2395,
      "step": 168
    },
    {
      "epoch": 0.21,
      "learning_rate": 4.8471971066907776e-05,
      "loss": 0.4398,
      "step": 169
    },
    {
      "epoch": 0.22,
      "learning_rate": 4.84629294755877e-05,
      "loss": 0.4757,
      "step": 170
    },
    {
      "epoch": 0.22,
      "learning_rate": 4.845388788426763e-05,
      "loss": 0.3461,
      "step": 171
    },
    {
      "epoch": 0.22,
      "learning_rate": 4.844484629294756e-05,
      "loss": 0.4196,
      "step": 172
    },
    {
      "epoch": 0.22,
      "learning_rate": 4.843580470162749e-05,
      "loss": 0.3559,
      "step": 173
    },
    {
      "epoch": 0.22,
      "learning_rate": 4.8426763110307414e-05,
      "loss": 0.3095,
      "step": 174
    },
    {
      "epoch": 0.22,
      "learning_rate": 4.8417721518987346e-05,
      "loss": 0.3509,
      "step": 175
    },
    {
      "epoch": 0.22,
      "learning_rate": 4.840867992766727e-05,
      "loss": 0.439,
      "step": 176
    },
    {
      "epoch": 0.22,
      "learning_rate": 4.8399638336347195e-05,
      "loss": 0.2752,
      "step": 177
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.839059674502713e-05,
      "loss": 0.4862,
      "step": 178
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.838155515370705e-05,
      "loss": 0.4599,
      "step": 179
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.8372513562386984e-05,
      "loss": 0.3442,
      "step": 180
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.836347197106691e-05,
      "loss": 0.2045,
      "step": 181
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.835443037974684e-05,
      "loss": 0.2513,
      "step": 182
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.8345388788426765e-05,
      "loss": 0.2156,
      "step": 183
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.833634719710669e-05,
      "loss": 0.4379,
      "step": 184
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.832730560578662e-05,
      "loss": 0.3841,
      "step": 185
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.8318264014466546e-05,
      "loss": 0.3266,
      "step": 186
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.830922242314648e-05,
      "loss": 0.3674,
      "step": 187
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.83001808318264e-05,
      "loss": 0.1755,
      "step": 188
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.8291139240506334e-05,
      "loss": 0.357,
      "step": 189
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.828209764918626e-05,
      "loss": 0.7705,
      "step": 190
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.827305605786619e-05,
      "loss": 0.5736,
      "step": 191
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.8264014466546116e-05,
      "loss": 0.2115,
      "step": 192
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.825497287522604e-05,
      "loss": 0.3475,
      "step": 193
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.824593128390597e-05,
      "loss": 0.3407,
      "step": 194
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.82368896925859e-05,
      "loss": 0.3339,
      "step": 195
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.822784810126583e-05,
      "loss": 0.4501,
      "step": 196
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.8218806509945753e-05,
      "loss": 0.3184,
      "step": 197
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.8209764918625685e-05,
      "loss": 0.3199,
      "step": 198
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.820072332730561e-05,
      "loss": 0.2649,
      "step": 199
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.8191681735985535e-05,
      "loss": 0.2321,
      "step": 200
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.8182640144665466e-05,
      "loss": 0.3216,
      "step": 201
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.817359855334539e-05,
      "loss": 0.3767,
      "step": 202
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.816455696202532e-05,
      "loss": 0.3347,
      "step": 203
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.815551537070525e-05,
      "loss": 0.1514,
      "step": 204
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.814647377938518e-05,
      "loss": 0.5632,
      "step": 205
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.8137432188065104e-05,
      "loss": 0.3482,
      "step": 206
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.812839059674503e-05,
      "loss": 0.3698,
      "step": 207
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.811934900542496e-05,
      "loss": 0.3587,
      "step": 208
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.8110307414104886e-05,
      "loss": 0.2163,
      "step": 209
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.810126582278481e-05,
      "loss": 0.1754,
      "step": 210
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.809222423146474e-05,
      "loss": 0.3963,
      "step": 211
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.808318264014467e-05,
      "loss": 0.1646,
      "step": 212
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.807414104882459e-05,
      "loss": 0.2231,
      "step": 213
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.806509945750452e-05,
      "loss": 0.3407,
      "step": 214
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.805605786618445e-05,
      "loss": 0.215,
      "step": 215
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.804701627486437e-05,
      "loss": 0.3084,
      "step": 216
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.8037974683544305e-05,
      "loss": 0.4503,
      "step": 217
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.802893309222423e-05,
      "loss": 0.2793,
      "step": 218
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.801989150090416e-05,
      "loss": 0.386,
      "step": 219
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.8010849909584086e-05,
      "loss": 0.3978,
      "step": 220
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.800180831826402e-05,
      "loss": 0.4766,
      "step": 221
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.799276672694394e-05,
      "loss": 0.2517,
      "step": 222
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.7983725135623874e-05,
      "loss": 0.2024,
      "step": 223
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.79746835443038e-05,
      "loss": 0.1937,
      "step": 224
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.7965641952983724e-05,
      "loss": 0.3569,
      "step": 225
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.7956600361663655e-05,
      "loss": 0.1182,
      "step": 226
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.794755877034358e-05,
      "loss": 0.2636,
      "step": 227
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.793851717902351e-05,
      "loss": 0.3986,
      "step": 228
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.792947558770344e-05,
      "loss": 0.3707,
      "step": 229
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.792043399638337e-05,
      "loss": 0.4346,
      "step": 230
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.791139240506329e-05,
      "loss": 0.1537,
      "step": 231
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.790235081374322e-05,
      "loss": 0.3636,
      "step": 232
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.789330922242315e-05,
      "loss": 0.2421,
      "step": 233
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.7884267631103075e-05,
      "loss": 0.3207,
      "step": 234
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.7875226039783006e-05,
      "loss": 0.1279,
      "step": 235
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.786618444846293e-05,
      "loss": 0.2593,
      "step": 236
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.785714285714286e-05,
      "loss": 0.4357,
      "step": 237
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.784810126582279e-05,
      "loss": 0.3013,
      "step": 238
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.783905967450272e-05,
      "loss": 0.329,
      "step": 239
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.7830018083182644e-05,
      "loss": 0.2702,
      "step": 240
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.782097649186257e-05,
      "loss": 0.1149,
      "step": 241
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.78119349005425e-05,
      "loss": 0.3434,
      "step": 242
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.7802893309222425e-05,
      "loss": 0.2747,
      "step": 243
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.779385171790236e-05,
      "loss": 0.4895,
      "step": 244
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.778481012658228e-05,
      "loss": 0.2141,
      "step": 245
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.777576853526221e-05,
      "loss": 0.2643,
      "step": 246
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.776672694394214e-05,
      "loss": 0.3403,
      "step": 247
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.775768535262206e-05,
      "loss": 0.2513,
      "step": 248
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.7748643761301995e-05,
      "loss": 0.2791,
      "step": 249
    },
    {
      "epoch": 0.32,
      "learning_rate": 4.773960216998192e-05,
      "loss": 0.3684,
      "step": 250
    },
    {
      "epoch": 0.32,
      "eval_accuracy": 0.8753753753753754,
      "eval_loss": 0.3046382665634155,
      "eval_runtime": 45.1323,
      "eval_samples_per_second": 29.513,
      "eval_steps_per_second": 3.7,
      "step": 250
    },
    {
      "epoch": 0.32,
      "learning_rate": 4.773056057866185e-05,
      "loss": 0.262,
      "step": 251
    },
    {
      "epoch": 0.32,
      "learning_rate": 4.7721518987341776e-05,
      "loss": 0.5526,
      "step": 252
    },
    {
      "epoch": 0.32,
      "learning_rate": 4.77124773960217e-05,
      "loss": 0.3508,
      "step": 253
    },
    {
      "epoch": 0.32,
      "learning_rate": 4.770343580470163e-05,
      "loss": 0.4899,
      "step": 254
    },
    {
      "epoch": 0.32,
      "learning_rate": 4.769439421338156e-05,
      "loss": 0.2026,
      "step": 255
    },
    {
      "epoch": 0.32,
      "learning_rate": 4.768535262206148e-05,
      "loss": 0.4545,
      "step": 256
    },
    {
      "epoch": 0.33,
      "learning_rate": 4.7676311030741414e-05,
      "loss": 0.2752,
      "step": 257
    },
    {
      "epoch": 0.33,
      "learning_rate": 4.766726943942134e-05,
      "loss": 0.472,
      "step": 258
    },
    {
      "epoch": 0.33,
      "learning_rate": 4.7658227848101264e-05,
      "loss": 0.44,
      "step": 259
    },
    {
      "epoch": 0.33,
      "learning_rate": 4.7649186256781195e-05,
      "loss": 0.2261,
      "step": 260
    },
    {
      "epoch": 0.33,
      "learning_rate": 4.764014466546112e-05,
      "loss": 0.2758,
      "step": 261
    },
    {
      "epoch": 0.33,
      "learning_rate": 4.763110307414105e-05,
      "loss": 0.2112,
      "step": 262
    },
    {
      "epoch": 0.33,
      "learning_rate": 4.7622061482820976e-05,
      "loss": 0.2849,
      "step": 263
    },
    {
      "epoch": 0.33,
      "learning_rate": 4.76130198915009e-05,
      "loss": 0.4166,
      "step": 264
    },
    {
      "epoch": 0.34,
      "learning_rate": 4.760397830018083e-05,
      "loss": 0.216,
      "step": 265
    },
    {
      "epoch": 0.34,
      "learning_rate": 4.759493670886076e-05,
      "loss": 0.4678,
      "step": 266
    },
    {
      "epoch": 0.34,
      "learning_rate": 4.758589511754069e-05,
      "loss": 0.3787,
      "step": 267
    },
    {
      "epoch": 0.34,
      "learning_rate": 4.7576853526220614e-05,
      "loss": 0.1444,
      "step": 268
    },
    {
      "epoch": 0.34,
      "learning_rate": 4.7567811934900546e-05,
      "loss": 0.332,
      "step": 269
    },
    {
      "epoch": 0.34,
      "learning_rate": 4.755877034358047e-05,
      "loss": 0.3063,
      "step": 270
    },
    {
      "epoch": 0.34,
      "learning_rate": 4.7549728752260396e-05,
      "loss": 0.5223,
      "step": 271
    },
    {
      "epoch": 0.34,
      "learning_rate": 4.754068716094033e-05,
      "loss": 0.1629,
      "step": 272
    },
    {
      "epoch": 0.35,
      "learning_rate": 4.753164556962025e-05,
      "loss": 0.2504,
      "step": 273
    },
    {
      "epoch": 0.35,
      "learning_rate": 4.7522603978300184e-05,
      "loss": 0.4073,
      "step": 274
    },
    {
      "epoch": 0.35,
      "learning_rate": 4.751356238698011e-05,
      "loss": 0.2715,
      "step": 275
    },
    {
      "epoch": 0.35,
      "learning_rate": 4.750452079566004e-05,
      "loss": 0.3011,
      "step": 276
    },
    {
      "epoch": 0.35,
      "learning_rate": 4.7495479204339965e-05,
      "loss": 0.4324,
      "step": 277
    },
    {
      "epoch": 0.35,
      "learning_rate": 4.74864376130199e-05,
      "loss": 0.254,
      "step": 278
    },
    {
      "epoch": 0.35,
      "learning_rate": 4.747739602169982e-05,
      "loss": 0.3187,
      "step": 279
    },
    {
      "epoch": 0.35,
      "learning_rate": 4.7468354430379746e-05,
      "loss": 0.4391,
      "step": 280
    },
    {
      "epoch": 0.36,
      "learning_rate": 4.745931283905968e-05,
      "loss": 0.2249,
      "step": 281
    },
    {
      "epoch": 0.36,
      "learning_rate": 4.74502712477396e-05,
      "loss": 0.3935,
      "step": 282
    },
    {
      "epoch": 0.36,
      "learning_rate": 4.7441229656419534e-05,
      "loss": 0.3202,
      "step": 283
    },
    {
      "epoch": 0.36,
      "learning_rate": 4.743218806509946e-05,
      "loss": 0.1962,
      "step": 284
    },
    {
      "epoch": 0.36,
      "learning_rate": 4.742314647377939e-05,
      "loss": 0.277,
      "step": 285
    },
    {
      "epoch": 0.36,
      "learning_rate": 4.7414104882459316e-05,
      "loss": 0.3519,
      "step": 286
    },
    {
      "epoch": 0.36,
      "learning_rate": 4.740506329113924e-05,
      "loss": 0.3388,
      "step": 287
    },
    {
      "epoch": 0.36,
      "learning_rate": 4.739602169981917e-05,
      "loss": 0.1162,
      "step": 288
    },
    {
      "epoch": 0.37,
      "learning_rate": 4.73869801084991e-05,
      "loss": 0.1759,
      "step": 289
    },
    {
      "epoch": 0.37,
      "learning_rate": 4.737793851717903e-05,
      "loss": 0.2307,
      "step": 290
    },
    {
      "epoch": 0.37,
      "learning_rate": 4.7368896925858954e-05,
      "loss": 0.4024,
      "step": 291
    },
    {
      "epoch": 0.37,
      "learning_rate": 4.7359855334538885e-05,
      "loss": 0.4341,
      "step": 292
    },
    {
      "epoch": 0.37,
      "learning_rate": 4.735081374321881e-05,
      "loss": 0.216,
      "step": 293
    },
    {
      "epoch": 0.37,
      "learning_rate": 4.734177215189874e-05,
      "loss": 0.2266,
      "step": 294
    },
    {
      "epoch": 0.37,
      "learning_rate": 4.7332730560578667e-05,
      "loss": 0.3129,
      "step": 295
    },
    {
      "epoch": 0.37,
      "learning_rate": 4.732368896925859e-05,
      "loss": 0.458,
      "step": 296
    },
    {
      "epoch": 0.38,
      "learning_rate": 4.731464737793852e-05,
      "loss": 0.3499,
      "step": 297
    },
    {
      "epoch": 0.38,
      "learning_rate": 4.730560578661845e-05,
      "loss": 0.2545,
      "step": 298
    },
    {
      "epoch": 0.38,
      "learning_rate": 4.729656419529838e-05,
      "loss": 0.2782,
      "step": 299
    },
    {
      "epoch": 0.38,
      "learning_rate": 4.7287522603978304e-05,
      "loss": 0.2506,
      "step": 300
    },
    {
      "epoch": 0.38,
      "learning_rate": 4.727848101265823e-05,
      "loss": 0.2972,
      "step": 301
    },
    {
      "epoch": 0.38,
      "learning_rate": 4.726943942133816e-05,
      "loss": 0.3468,
      "step": 302
    },
    {
      "epoch": 0.38,
      "learning_rate": 4.7260397830018086e-05,
      "loss": 0.2567,
      "step": 303
    },
    {
      "epoch": 0.38,
      "learning_rate": 4.725135623869801e-05,
      "loss": 0.2996,
      "step": 304
    },
    {
      "epoch": 0.39,
      "learning_rate": 4.724231464737794e-05,
      "loss": 0.3314,
      "step": 305
    },
    {
      "epoch": 0.39,
      "learning_rate": 4.723327305605787e-05,
      "loss": 0.2544,
      "step": 306
    },
    {
      "epoch": 0.39,
      "learning_rate": 4.722423146473779e-05,
      "loss": 0.2551,
      "step": 307
    },
    {
      "epoch": 0.39,
      "learning_rate": 4.7215189873417723e-05,
      "loss": 0.2201,
      "step": 308
    },
    {
      "epoch": 0.39,
      "learning_rate": 4.720614828209765e-05,
      "loss": 0.334,
      "step": 309
    },
    {
      "epoch": 0.39,
      "learning_rate": 4.719710669077758e-05,
      "loss": 0.2678,
      "step": 310
    },
    {
      "epoch": 0.39,
      "learning_rate": 4.7188065099457505e-05,
      "loss": 0.272,
      "step": 311
    },
    {
      "epoch": 0.39,
      "learning_rate": 4.717902350813743e-05,
      "loss": 0.412,
      "step": 312
    },
    {
      "epoch": 0.4,
      "learning_rate": 4.716998191681736e-05,
      "loss": 0.4929,
      "step": 313
    },
    {
      "epoch": 0.4,
      "learning_rate": 4.7160940325497286e-05,
      "loss": 0.1627,
      "step": 314
    },
    {
      "epoch": 0.4,
      "learning_rate": 4.715189873417722e-05,
      "loss": 0.3099,
      "step": 315
    },
    {
      "epoch": 0.4,
      "learning_rate": 4.714285714285714e-05,
      "loss": 0.2658,
      "step": 316
    },
    {
      "epoch": 0.4,
      "learning_rate": 4.7133815551537074e-05,
      "loss": 0.3195,
      "step": 317
    },
    {
      "epoch": 0.4,
      "learning_rate": 4.7124773960217e-05,
      "loss": 0.4028,
      "step": 318
    },
    {
      "epoch": 0.4,
      "learning_rate": 4.7115732368896924e-05,
      "loss": 0.2019,
      "step": 319
    },
    {
      "epoch": 0.4,
      "learning_rate": 4.7106690777576856e-05,
      "loss": 0.3913,
      "step": 320
    },
    {
      "epoch": 0.41,
      "learning_rate": 4.709764918625678e-05,
      "loss": 0.1756,
      "step": 321
    },
    {
      "epoch": 0.41,
      "learning_rate": 4.708860759493671e-05,
      "loss": 0.3993,
      "step": 322
    },
    {
      "epoch": 0.41,
      "learning_rate": 4.707956600361664e-05,
      "loss": 0.2569,
      "step": 323
    },
    {
      "epoch": 0.41,
      "learning_rate": 4.707052441229657e-05,
      "loss": 0.1487,
      "step": 324
    },
    {
      "epoch": 0.41,
      "learning_rate": 4.706148282097649e-05,
      "loss": 0.2761,
      "step": 325
    },
    {
      "epoch": 0.41,
      "learning_rate": 4.7052441229656425e-05,
      "loss": 0.1945,
      "step": 326
    },
    {
      "epoch": 0.41,
      "learning_rate": 4.704339963833635e-05,
      "loss": 0.331,
      "step": 327
    },
    {
      "epoch": 0.41,
      "learning_rate": 4.7034358047016275e-05,
      "loss": 0.3684,
      "step": 328
    },
    {
      "epoch": 0.42,
      "learning_rate": 4.7025316455696206e-05,
      "loss": 0.2004,
      "step": 329
    },
    {
      "epoch": 0.42,
      "learning_rate": 4.701627486437613e-05,
      "loss": 0.2493,
      "step": 330
    },
    {
      "epoch": 0.42,
      "learning_rate": 4.700723327305606e-05,
      "loss": 0.3097,
      "step": 331
    },
    {
      "epoch": 0.42,
      "learning_rate": 4.699819168173599e-05,
      "loss": 0.3176,
      "step": 332
    },
    {
      "epoch": 0.42,
      "learning_rate": 4.698915009041592e-05,
      "loss": 0.2182,
      "step": 333
    },
    {
      "epoch": 0.42,
      "learning_rate": 4.6980108499095844e-05,
      "loss": 0.3965,
      "step": 334
    },
    {
      "epoch": 0.42,
      "learning_rate": 4.697106690777577e-05,
      "loss": 0.1666,
      "step": 335
    },
    {
      "epoch": 0.43,
      "learning_rate": 4.69620253164557e-05,
      "loss": 0.2218,
      "step": 336
    },
    {
      "epoch": 0.43,
      "learning_rate": 4.6952983725135625e-05,
      "loss": 0.126,
      "step": 337
    },
    {
      "epoch": 0.43,
      "learning_rate": 4.694394213381556e-05,
      "loss": 0.2831,
      "step": 338
    },
    {
      "epoch": 0.43,
      "learning_rate": 4.693490054249548e-05,
      "loss": 0.1907,
      "step": 339
    },
    {
      "epoch": 0.43,
      "learning_rate": 4.6925858951175414e-05,
      "loss": 0.3623,
      "step": 340
    },
    {
      "epoch": 0.43,
      "learning_rate": 4.691681735985534e-05,
      "loss": 0.3561,
      "step": 341
    },
    {
      "epoch": 0.43,
      "learning_rate": 4.690777576853527e-05,
      "loss": 0.4864,
      "step": 342
    },
    {
      "epoch": 0.43,
      "learning_rate": 4.6898734177215195e-05,
      "loss": 0.3366,
      "step": 343
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.688969258589512e-05,
      "loss": 0.2613,
      "step": 344
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.688065099457505e-05,
      "loss": 0.4724,
      "step": 345
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.6871609403254976e-05,
      "loss": 0.4303,
      "step": 346
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.68625678119349e-05,
      "loss": 0.1626,
      "step": 347
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.685352622061483e-05,
      "loss": 0.244,
      "step": 348
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.684448462929476e-05,
      "loss": 0.2372,
      "step": 349
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.683544303797468e-05,
      "loss": 0.2752,
      "step": 350
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.6826401446654614e-05,
      "loss": 0.1917,
      "step": 351
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.681735985533454e-05,
      "loss": 0.4126,
      "step": 352
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.6808318264014464e-05,
      "loss": 0.219,
      "step": 353
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.6799276672694395e-05,
      "loss": 0.268,
      "step": 354
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.679023508137432e-05,
      "loss": 0.1729,
      "step": 355
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.678119349005425e-05,
      "loss": 0.3509,
      "step": 356
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.677215189873418e-05,
      "loss": 0.2411,
      "step": 357
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.676311030741411e-05,
      "loss": 0.2342,
      "step": 358
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.675406871609403e-05,
      "loss": 0.118,
      "step": 359
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.674502712477396e-05,
      "loss": 0.4844,
      "step": 360
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.673598553345389e-05,
      "loss": 0.3585,
      "step": 361
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.6726943942133814e-05,
      "loss": 0.2128,
      "step": 362
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.6717902350813746e-05,
      "loss": 0.1477,
      "step": 363
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.670886075949367e-05,
      "loss": 0.3721,
      "step": 364
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.66998191681736e-05,
      "loss": 0.4461,
      "step": 365
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.669077757685353e-05,
      "loss": 0.192,
      "step": 366
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.668173598553345e-05,
      "loss": 0.1981,
      "step": 367
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.6672694394213384e-05,
      "loss": 0.2047,
      "step": 368
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.666365280289331e-05,
      "loss": 0.2673,
      "step": 369
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.665461121157324e-05,
      "loss": 0.2521,
      "step": 370
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.6645569620253165e-05,
      "loss": 0.208,
      "step": 371
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.66365280289331e-05,
      "loss": 0.1358,
      "step": 372
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.662748643761302e-05,
      "loss": 0.304,
      "step": 373
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.661844484629295e-05,
      "loss": 0.4907,
      "step": 374
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.660940325497288e-05,
      "loss": 0.191,
      "step": 375
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.66003616636528e-05,
      "loss": 0.1937,
      "step": 376
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.6591320072332735e-05,
      "loss": 0.339,
      "step": 377
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.658227848101266e-05,
      "loss": 0.2427,
      "step": 378
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.657323688969259e-05,
      "loss": 0.3593,
      "step": 379
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.6564195298372516e-05,
      "loss": 0.3804,
      "step": 380
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.655515370705245e-05,
      "loss": 0.4035,
      "step": 381
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.654611211573237e-05,
      "loss": 0.4135,
      "step": 382
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.65370705244123e-05,
      "loss": 0.1407,
      "step": 383
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.652802893309223e-05,
      "loss": 0.3355,
      "step": 384
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.6518987341772154e-05,
      "loss": 0.4381,
      "step": 385
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.6509945750452085e-05,
      "loss": 0.4553,
      "step": 386
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.650090415913201e-05,
      "loss": 0.2707,
      "step": 387
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.649186256781194e-05,
      "loss": 0.3847,
      "step": 388
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.648282097649187e-05,
      "loss": 0.2034,
      "step": 389
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.64737793851718e-05,
      "loss": 0.2079,
      "step": 390
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.646473779385172e-05,
      "loss": 0.2128,
      "step": 391
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.645569620253165e-05,
      "loss": 0.2942,
      "step": 392
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.644665461121158e-05,
      "loss": 0.1548,
      "step": 393
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.6437613019891504e-05,
      "loss": 0.3921,
      "step": 394
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.642857142857143e-05,
      "loss": 0.3494,
      "step": 395
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.641952983725136e-05,
      "loss": 0.2422,
      "step": 396
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.6410488245931286e-05,
      "loss": 0.3686,
      "step": 397
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.640144665461121e-05,
      "loss": 0.5112,
      "step": 398
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.6392405063291136e-05,
      "loss": 0.3493,
      "step": 399
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.638336347197107e-05,
      "loss": 0.2968,
      "step": 400
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.637432188065099e-05,
      "loss": 0.2745,
      "step": 401
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.6365280289330924e-05,
      "loss": 0.2469,
      "step": 402
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.635623869801085e-05,
      "loss": 0.2392,
      "step": 403
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.634719710669078e-05,
      "loss": 0.3077,
      "step": 404
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.6338155515370705e-05,
      "loss": 0.5113,
      "step": 405
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.6329113924050637e-05,
      "loss": 0.1666,
      "step": 406
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.632007233273056e-05,
      "loss": 0.1338,
      "step": 407
    },
    {
      "epoch": 0.52,
      "learning_rate": 4.6311030741410486e-05,
      "loss": 0.3951,
      "step": 408
    },
    {
      "epoch": 0.52,
      "learning_rate": 4.630198915009042e-05,
      "loss": 0.2627,
      "step": 409
    },
    {
      "epoch": 0.52,
      "learning_rate": 4.629294755877034e-05,
      "loss": 0.142,
      "step": 410
    },
    {
      "epoch": 0.52,
      "learning_rate": 4.6283905967450274e-05,
      "loss": 0.2112,
      "step": 411
    },
    {
      "epoch": 0.52,
      "learning_rate": 4.62748643761302e-05,
      "loss": 0.2092,
      "step": 412
    },
    {
      "epoch": 0.52,
      "learning_rate": 4.626582278481013e-05,
      "loss": 0.2319,
      "step": 413
    },
    {
      "epoch": 0.52,
      "learning_rate": 4.6256781193490056e-05,
      "loss": 0.2885,
      "step": 414
    },
    {
      "epoch": 0.52,
      "learning_rate": 4.624773960216998e-05,
      "loss": 0.2227,
      "step": 415
    },
    {
      "epoch": 0.53,
      "learning_rate": 4.623869801084991e-05,
      "loss": 0.2235,
      "step": 416
    },
    {
      "epoch": 0.53,
      "learning_rate": 4.622965641952984e-05,
      "loss": 0.4433,
      "step": 417
    },
    {
      "epoch": 0.53,
      "learning_rate": 4.622061482820977e-05,
      "loss": 0.1563,
      "step": 418
    },
    {
      "epoch": 0.53,
      "learning_rate": 4.6211573236889694e-05,
      "loss": 0.4165,
      "step": 419
    },
    {
      "epoch": 0.53,
      "learning_rate": 4.6202531645569625e-05,
      "loss": 0.2846,
      "step": 420
    },
    {
      "epoch": 0.53,
      "learning_rate": 4.619349005424955e-05,
      "loss": 0.27,
      "step": 421
    },
    {
      "epoch": 0.53,
      "learning_rate": 4.618444846292948e-05,
      "loss": 0.4965,
      "step": 422
    },
    {
      "epoch": 0.54,
      "learning_rate": 4.6175406871609406e-05,
      "loss": 0.1533,
      "step": 423
    },
    {
      "epoch": 0.54,
      "learning_rate": 4.616636528028933e-05,
      "loss": 0.112,
      "step": 424
    },
    {
      "epoch": 0.54,
      "learning_rate": 4.615732368896926e-05,
      "loss": 0.2695,
      "step": 425
    },
    {
      "epoch": 0.54,
      "learning_rate": 4.614828209764919e-05,
      "loss": 0.2582,
      "step": 426
    },
    {
      "epoch": 0.54,
      "learning_rate": 4.613924050632912e-05,
      "loss": 0.2821,
      "step": 427
    },
    {
      "epoch": 0.54,
      "learning_rate": 4.6130198915009044e-05,
      "loss": 0.3077,
      "step": 428
    },
    {
      "epoch": 0.54,
      "learning_rate": 4.6121157323688976e-05,
      "loss": 0.1766,
      "step": 429
    },
    {
      "epoch": 0.54,
      "learning_rate": 4.61121157323689e-05,
      "loss": 0.1456,
      "step": 430
    },
    {
      "epoch": 0.55,
      "learning_rate": 4.6103074141048826e-05,
      "loss": 0.1668,
      "step": 431
    },
    {
      "epoch": 0.55,
      "learning_rate": 4.609403254972876e-05,
      "loss": 0.2006,
      "step": 432
    },
    {
      "epoch": 0.55,
      "learning_rate": 4.608499095840868e-05,
      "loss": 0.1975,
      "step": 433
    },
    {
      "epoch": 0.55,
      "learning_rate": 4.6075949367088614e-05,
      "loss": 0.2004,
      "step": 434
    },
    {
      "epoch": 0.55,
      "learning_rate": 4.606690777576854e-05,
      "loss": 0.3774,
      "step": 435
    },
    {
      "epoch": 0.55,
      "learning_rate": 4.605786618444847e-05,
      "loss": 0.2323,
      "step": 436
    },
    {
      "epoch": 0.55,
      "learning_rate": 4.6048824593128395e-05,
      "loss": 0.3245,
      "step": 437
    },
    {
      "epoch": 0.55,
      "learning_rate": 4.603978300180832e-05,
      "loss": 0.2687,
      "step": 438
    },
    {
      "epoch": 0.56,
      "learning_rate": 4.603074141048825e-05,
      "loss": 0.4587,
      "step": 439
    },
    {
      "epoch": 0.56,
      "learning_rate": 4.6021699819168176e-05,
      "loss": 0.3295,
      "step": 440
    },
    {
      "epoch": 0.56,
      "learning_rate": 4.60126582278481e-05,
      "loss": 0.2194,
      "step": 441
    },
    {
      "epoch": 0.56,
      "learning_rate": 4.600361663652803e-05,
      "loss": 0.1599,
      "step": 442
    },
    {
      "epoch": 0.56,
      "learning_rate": 4.599457504520796e-05,
      "loss": 0.2929,
      "step": 443
    },
    {
      "epoch": 0.56,
      "learning_rate": 4.598553345388788e-05,
      "loss": 0.2395,
      "step": 444
    },
    {
      "epoch": 0.56,
      "learning_rate": 4.5976491862567814e-05,
      "loss": 0.3794,
      "step": 445
    },
    {
      "epoch": 0.56,
      "learning_rate": 4.596745027124774e-05,
      "loss": 0.3383,
      "step": 446
    },
    {
      "epoch": 0.57,
      "learning_rate": 4.5958408679927664e-05,
      "loss": 0.2934,
      "step": 447
    },
    {
      "epoch": 0.57,
      "learning_rate": 4.5949367088607595e-05,
      "loss": 0.3143,
      "step": 448
    },
    {
      "epoch": 0.57,
      "learning_rate": 4.594032549728752e-05,
      "loss": 0.1366,
      "step": 449
    },
    {
      "epoch": 0.57,
      "learning_rate": 4.593128390596745e-05,
      "loss": 0.3903,
      "step": 450
    },
    {
      "epoch": 0.57,
      "learning_rate": 4.592224231464738e-05,
      "loss": 0.3288,
      "step": 451
    },
    {
      "epoch": 0.57,
      "learning_rate": 4.591320072332731e-05,
      "loss": 0.2494,
      "step": 452
    },
    {
      "epoch": 0.57,
      "learning_rate": 4.590415913200723e-05,
      "loss": 0.2568,
      "step": 453
    },
    {
      "epoch": 0.57,
      "learning_rate": 4.5895117540687165e-05,
      "loss": 0.281,
      "step": 454
    },
    {
      "epoch": 0.58,
      "learning_rate": 4.588607594936709e-05,
      "loss": 0.1467,
      "step": 455
    },
    {
      "epoch": 0.58,
      "learning_rate": 4.5877034358047015e-05,
      "loss": 0.383,
      "step": 456
    },
    {
      "epoch": 0.58,
      "learning_rate": 4.5867992766726946e-05,
      "loss": 0.1558,
      "step": 457
    },
    {
      "epoch": 0.58,
      "learning_rate": 4.585895117540687e-05,
      "loss": 0.3054,
      "step": 458
    },
    {
      "epoch": 0.58,
      "learning_rate": 4.58499095840868e-05,
      "loss": 0.1825,
      "step": 459
    },
    {
      "epoch": 0.58,
      "learning_rate": 4.584086799276673e-05,
      "loss": 0.4455,
      "step": 460
    },
    {
      "epoch": 0.58,
      "learning_rate": 4.583182640144666e-05,
      "loss": 0.5582,
      "step": 461
    },
    {
      "epoch": 0.58,
      "learning_rate": 4.5822784810126584e-05,
      "loss": 0.5069,
      "step": 462
    },
    {
      "epoch": 0.59,
      "learning_rate": 4.581374321880651e-05,
      "loss": 0.3907,
      "step": 463
    },
    {
      "epoch": 0.59,
      "learning_rate": 4.580470162748644e-05,
      "loss": 0.309,
      "step": 464
    },
    {
      "epoch": 0.59,
      "learning_rate": 4.5795660036166365e-05,
      "loss": 0.2404,
      "step": 465
    },
    {
      "epoch": 0.59,
      "learning_rate": 4.57866184448463e-05,
      "loss": 0.2825,
      "step": 466
    },
    {
      "epoch": 0.59,
      "learning_rate": 4.577757685352622e-05,
      "loss": 0.2451,
      "step": 467
    },
    {
      "epoch": 0.59,
      "learning_rate": 4.5768535262206153e-05,
      "loss": 0.1388,
      "step": 468
    },
    {
      "epoch": 0.59,
      "learning_rate": 4.575949367088608e-05,
      "loss": 0.1821,
      "step": 469
    },
    {
      "epoch": 0.59,
      "learning_rate": 4.575045207956601e-05,
      "loss": 0.2192,
      "step": 470
    },
    {
      "epoch": 0.6,
      "learning_rate": 4.5741410488245935e-05,
      "loss": 0.3942,
      "step": 471
    },
    {
      "epoch": 0.6,
      "learning_rate": 4.573236889692586e-05,
      "loss": 0.1332,
      "step": 472
    },
    {
      "epoch": 0.6,
      "learning_rate": 4.572332730560579e-05,
      "loss": 0.5592,
      "step": 473
    },
    {
      "epoch": 0.6,
      "learning_rate": 4.5714285714285716e-05,
      "loss": 0.1815,
      "step": 474
    },
    {
      "epoch": 0.6,
      "learning_rate": 4.570524412296565e-05,
      "loss": 0.1509,
      "step": 475
    },
    {
      "epoch": 0.6,
      "learning_rate": 4.569620253164557e-05,
      "loss": 0.2022,
      "step": 476
    },
    {
      "epoch": 0.6,
      "learning_rate": 4.5687160940325504e-05,
      "loss": 0.2739,
      "step": 477
    },
    {
      "epoch": 0.6,
      "learning_rate": 4.567811934900543e-05,
      "loss": 0.1964,
      "step": 478
    },
    {
      "epoch": 0.61,
      "learning_rate": 4.5669077757685354e-05,
      "loss": 0.2264,
      "step": 479
    },
    {
      "epoch": 0.61,
      "learning_rate": 4.5660036166365286e-05,
      "loss": 0.1262,
      "step": 480
    },
    {
      "epoch": 0.61,
      "learning_rate": 4.565099457504521e-05,
      "loss": 0.3119,
      "step": 481
    },
    {
      "epoch": 0.61,
      "learning_rate": 4.564195298372514e-05,
      "loss": 0.3087,
      "step": 482
    },
    {
      "epoch": 0.61,
      "learning_rate": 4.563291139240507e-05,
      "loss": 0.1936,
      "step": 483
    },
    {
      "epoch": 0.61,
      "learning_rate": 4.5623869801085e-05,
      "loss": 0.2896,
      "step": 484
    },
    {
      "epoch": 0.61,
      "learning_rate": 4.561482820976492e-05,
      "loss": 0.2228,
      "step": 485
    },
    {
      "epoch": 0.61,
      "learning_rate": 4.560578661844485e-05,
      "loss": 0.6279,
      "step": 486
    },
    {
      "epoch": 0.62,
      "learning_rate": 4.559674502712478e-05,
      "loss": 0.4198,
      "step": 487
    },
    {
      "epoch": 0.62,
      "learning_rate": 4.5587703435804705e-05,
      "loss": 0.2069,
      "step": 488
    },
    {
      "epoch": 0.62,
      "learning_rate": 4.557866184448463e-05,
      "loss": 0.2215,
      "step": 489
    },
    {
      "epoch": 0.62,
      "learning_rate": 4.556962025316456e-05,
      "loss": 0.1135,
      "step": 490
    },
    {
      "epoch": 0.62,
      "learning_rate": 4.5560578661844486e-05,
      "loss": 0.2719,
      "step": 491
    },
    {
      "epoch": 0.62,
      "learning_rate": 4.555153707052441e-05,
      "loss": 0.4347,
      "step": 492
    },
    {
      "epoch": 0.62,
      "learning_rate": 4.554249547920434e-05,
      "loss": 0.3283,
      "step": 493
    },
    {
      "epoch": 0.62,
      "learning_rate": 4.553345388788427e-05,
      "loss": 0.235,
      "step": 494
    },
    {
      "epoch": 0.63,
      "learning_rate": 4.552441229656419e-05,
      "loss": 0.4858,
      "step": 495
    },
    {
      "epoch": 0.63,
      "learning_rate": 4.5515370705244124e-05,
      "loss": 0.3099,
      "step": 496
    },
    {
      "epoch": 0.63,
      "learning_rate": 4.550632911392405e-05,
      "loss": 0.4075,
      "step": 497
    },
    {
      "epoch": 0.63,
      "learning_rate": 4.549728752260398e-05,
      "loss": 0.1918,
      "step": 498
    },
    {
      "epoch": 0.63,
      "learning_rate": 4.5488245931283905e-05,
      "loss": 0.1662,
      "step": 499
    },
    {
      "epoch": 0.63,
      "learning_rate": 4.547920433996384e-05,
      "loss": 0.3516,
      "step": 500
    },
    {
      "epoch": 0.63,
      "eval_accuracy": 0.9046546546546547,
      "eval_loss": 0.25014105439186096,
      "eval_runtime": 45.1236,
      "eval_samples_per_second": 29.519,
      "eval_steps_per_second": 3.701,
      "step": 500
    },
    {
      "epoch": 0.63,
      "learning_rate": 4.547016274864376e-05,
      "loss": 0.2568,
      "step": 501
    },
    {
      "epoch": 0.64,
      "learning_rate": 4.546112115732369e-05,
      "loss": 0.2309,
      "step": 502
    },
    {
      "epoch": 0.64,
      "learning_rate": 4.545207956600362e-05,
      "loss": 0.4003,
      "step": 503
    },
    {
      "epoch": 0.64,
      "learning_rate": 4.544303797468354e-05,
      "loss": 0.1692,
      "step": 504
    },
    {
      "epoch": 0.64,
      "learning_rate": 4.5433996383363475e-05,
      "loss": 0.229,
      "step": 505
    },
    {
      "epoch": 0.64,
      "learning_rate": 4.54249547920434e-05,
      "loss": 0.37,
      "step": 506
    },
    {
      "epoch": 0.64,
      "learning_rate": 4.541591320072333e-05,
      "loss": 0.2501,
      "step": 507
    },
    {
      "epoch": 0.64,
      "learning_rate": 4.5406871609403256e-05,
      "loss": 0.4087,
      "step": 508
    },
    {
      "epoch": 0.64,
      "learning_rate": 4.539783001808319e-05,
      "loss": 0.3517,
      "step": 509
    },
    {
      "epoch": 0.65,
      "learning_rate": 4.538878842676311e-05,
      "loss": 0.2593,
      "step": 510
    },
    {
      "epoch": 0.65,
      "learning_rate": 4.537974683544304e-05,
      "loss": 0.096,
      "step": 511
    },
    {
      "epoch": 0.65,
      "learning_rate": 4.537070524412297e-05,
      "loss": 0.1463,
      "step": 512
    },
    {
      "epoch": 0.65,
      "learning_rate": 4.5361663652802894e-05,
      "loss": 0.2692,
      "step": 513
    },
    {
      "epoch": 0.65,
      "learning_rate": 4.5352622061482825e-05,
      "loss": 0.2327,
      "step": 514
    },
    {
      "epoch": 0.65,
      "learning_rate": 4.534358047016275e-05,
      "loss": 0.3396,
      "step": 515
    },
    {
      "epoch": 0.65,
      "learning_rate": 4.533453887884268e-05,
      "loss": 0.2,
      "step": 516
    },
    {
      "epoch": 0.65,
      "learning_rate": 4.5325497287522607e-05,
      "loss": 0.2079,
      "step": 517
    },
    {
      "epoch": 0.66,
      "learning_rate": 4.531645569620253e-05,
      "loss": 0.259,
      "step": 518
    },
    {
      "epoch": 0.66,
      "learning_rate": 4.530741410488246e-05,
      "loss": 0.3419,
      "step": 519
    },
    {
      "epoch": 0.66,
      "learning_rate": 4.529837251356239e-05,
      "loss": 0.2436,
      "step": 520
    },
    {
      "epoch": 0.66,
      "learning_rate": 4.528933092224232e-05,
      "loss": 0.1192,
      "step": 521
    },
    {
      "epoch": 0.66,
      "learning_rate": 4.5280289330922244e-05,
      "loss": 0.239,
      "step": 522
    },
    {
      "epoch": 0.66,
      "learning_rate": 4.5271247739602176e-05,
      "loss": 0.1103,
      "step": 523
    },
    {
      "epoch": 0.66,
      "learning_rate": 4.52622061482821e-05,
      "loss": 0.1896,
      "step": 524
    },
    {
      "epoch": 0.66,
      "learning_rate": 4.525316455696203e-05,
      "loss": 0.3273,
      "step": 525
    },
    {
      "epoch": 0.67,
      "learning_rate": 4.524412296564196e-05,
      "loss": 0.3502,
      "step": 526
    },
    {
      "epoch": 0.67,
      "learning_rate": 4.523508137432188e-05,
      "loss": 0.1623,
      "step": 527
    },
    {
      "epoch": 0.67,
      "learning_rate": 4.5226039783001814e-05,
      "loss": 0.2391,
      "step": 528
    },
    {
      "epoch": 0.67,
      "learning_rate": 4.521699819168174e-05,
      "loss": 0.1211,
      "step": 529
    },
    {
      "epoch": 0.67,
      "learning_rate": 4.520795660036167e-05,
      "loss": 0.3811,
      "step": 530
    },
    {
      "epoch": 0.67,
      "learning_rate": 4.5198915009041595e-05,
      "loss": 0.1661,
      "step": 531
    },
    {
      "epoch": 0.67,
      "learning_rate": 4.518987341772152e-05,
      "loss": 0.1186,
      "step": 532
    },
    {
      "epoch": 0.67,
      "learning_rate": 4.518083182640145e-05,
      "loss": 0.3164,
      "step": 533
    },
    {
      "epoch": 0.68,
      "learning_rate": 4.5171790235081376e-05,
      "loss": 0.4179,
      "step": 534
    },
    {
      "epoch": 0.68,
      "learning_rate": 4.51627486437613e-05,
      "loss": 0.1748,
      "step": 535
    },
    {
      "epoch": 0.68,
      "learning_rate": 4.515370705244123e-05,
      "loss": 0.3072,
      "step": 536
    },
    {
      "epoch": 0.68,
      "learning_rate": 4.514466546112116e-05,
      "loss": 0.238,
      "step": 537
    },
    {
      "epoch": 0.68,
      "learning_rate": 4.513562386980108e-05,
      "loss": 0.2639,
      "step": 538
    },
    {
      "epoch": 0.68,
      "learning_rate": 4.5126582278481014e-05,
      "loss": 0.2392,
      "step": 539
    },
    {
      "epoch": 0.68,
      "learning_rate": 4.511754068716094e-05,
      "loss": 0.2617,
      "step": 540
    },
    {
      "epoch": 0.68,
      "learning_rate": 4.510849909584087e-05,
      "loss": 0.2701,
      "step": 541
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.5099457504520796e-05,
      "loss": 0.214,
      "step": 542
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.509041591320072e-05,
      "loss": 0.2611,
      "step": 543
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.508137432188065e-05,
      "loss": 0.2224,
      "step": 544
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.507233273056058e-05,
      "loss": 0.157,
      "step": 545
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.506329113924051e-05,
      "loss": 0.1501,
      "step": 546
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.5054249547920433e-05,
      "loss": 0.2033,
      "step": 547
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.5045207956600365e-05,
      "loss": 0.3791,
      "step": 548
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.503616636528029e-05,
      "loss": 0.3309,
      "step": 549
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.5027124773960215e-05,
      "loss": 0.4436,
      "step": 550
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.5018083182640146e-05,
      "loss": 0.2581,
      "step": 551
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.500904159132007e-05,
      "loss": 0.5388,
      "step": 552
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.5e-05,
      "loss": 0.0663,
      "step": 553
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.499095840867993e-05,
      "loss": 0.2075,
      "step": 554
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.498191681735986e-05,
      "loss": 0.2534,
      "step": 555
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.4972875226039784e-05,
      "loss": 0.3767,
      "step": 556
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.4963833634719716e-05,
      "loss": 0.1809,
      "step": 557
    },
    {
      "epoch": 0.71,
      "learning_rate": 4.495479204339964e-05,
      "loss": 0.3112,
      "step": 558
    },
    {
      "epoch": 0.71,
      "learning_rate": 4.4945750452079565e-05,
      "loss": 0.287,
      "step": 559
    },
    {
      "epoch": 0.71,
      "learning_rate": 4.49367088607595e-05,
      "loss": 0.4397,
      "step": 560
    },
    {
      "epoch": 0.71,
      "learning_rate": 4.492766726943942e-05,
      "loss": 0.3525,
      "step": 561
    },
    {
      "epoch": 0.71,
      "learning_rate": 4.4918625678119354e-05,
      "loss": 0.2468,
      "step": 562
    },
    {
      "epoch": 0.71,
      "learning_rate": 4.490958408679928e-05,
      "loss": 0.155,
      "step": 563
    },
    {
      "epoch": 0.71,
      "learning_rate": 4.490054249547921e-05,
      "loss": 0.2477,
      "step": 564
    },
    {
      "epoch": 0.71,
      "learning_rate": 4.4891500904159135e-05,
      "loss": 0.3951,
      "step": 565
    },
    {
      "epoch": 0.72,
      "learning_rate": 4.488245931283906e-05,
      "loss": 0.3133,
      "step": 566
    },
    {
      "epoch": 0.72,
      "learning_rate": 4.487341772151899e-05,
      "loss": 0.1769,
      "step": 567
    },
    {
      "epoch": 0.72,
      "learning_rate": 4.4864376130198916e-05,
      "loss": 0.2238,
      "step": 568
    },
    {
      "epoch": 0.72,
      "learning_rate": 4.485533453887885e-05,
      "loss": 0.2413,
      "step": 569
    },
    {
      "epoch": 0.72,
      "learning_rate": 4.484629294755877e-05,
      "loss": 0.1045,
      "step": 570
    },
    {
      "epoch": 0.72,
      "learning_rate": 4.4837251356238704e-05,
      "loss": 0.2755,
      "step": 571
    },
    {
      "epoch": 0.72,
      "learning_rate": 4.482820976491863e-05,
      "loss": 0.3366,
      "step": 572
    },
    {
      "epoch": 0.72,
      "learning_rate": 4.481916817359856e-05,
      "loss": 0.4702,
      "step": 573
    },
    {
      "epoch": 0.73,
      "learning_rate": 4.4810126582278486e-05,
      "loss": 0.2764,
      "step": 574
    },
    {
      "epoch": 0.73,
      "learning_rate": 4.480108499095841e-05,
      "loss": 0.1281,
      "step": 575
    },
    {
      "epoch": 0.73,
      "learning_rate": 4.479204339963834e-05,
      "loss": 0.1775,
      "step": 576
    },
    {
      "epoch": 0.73,
      "learning_rate": 4.478300180831827e-05,
      "loss": 0.2448,
      "step": 577
    },
    {
      "epoch": 0.73,
      "learning_rate": 4.47739602169982e-05,
      "loss": 0.1632,
      "step": 578
    },
    {
      "epoch": 0.73,
      "learning_rate": 4.4764918625678123e-05,
      "loss": 0.3721,
      "step": 579
    },
    {
      "epoch": 0.73,
      "learning_rate": 4.475587703435805e-05,
      "loss": 0.2953,
      "step": 580
    },
    {
      "epoch": 0.73,
      "learning_rate": 4.474683544303798e-05,
      "loss": 0.1839,
      "step": 581
    },
    {
      "epoch": 0.74,
      "learning_rate": 4.4737793851717905e-05,
      "loss": 0.3358,
      "step": 582
    },
    {
      "epoch": 0.74,
      "learning_rate": 4.472875226039783e-05,
      "loss": 0.2022,
      "step": 583
    },
    {
      "epoch": 0.74,
      "learning_rate": 4.4719710669077754e-05,
      "loss": 0.1676,
      "step": 584
    },
    {
      "epoch": 0.74,
      "learning_rate": 4.4710669077757686e-05,
      "loss": 0.2809,
      "step": 585
    },
    {
      "epoch": 0.74,
      "learning_rate": 4.470162748643761e-05,
      "loss": 0.2845,
      "step": 586
    },
    {
      "epoch": 0.74,
      "learning_rate": 4.469258589511754e-05,
      "loss": 0.1883,
      "step": 587
    },
    {
      "epoch": 0.74,
      "learning_rate": 4.468354430379747e-05,
      "loss": 0.3024,
      "step": 588
    },
    {
      "epoch": 0.75,
      "learning_rate": 4.46745027124774e-05,
      "loss": 0.1772,
      "step": 589
    },
    {
      "epoch": 0.75,
      "learning_rate": 4.4665461121157324e-05,
      "loss": 0.1234,
      "step": 590
    },
    {
      "epoch": 0.75,
      "learning_rate": 4.465641952983725e-05,
      "loss": 0.4229,
      "step": 591
    },
    {
      "epoch": 0.75,
      "learning_rate": 4.464737793851718e-05,
      "loss": 0.3691,
      "step": 592
    },
    {
      "epoch": 0.75,
      "learning_rate": 4.4638336347197105e-05,
      "loss": 0.5938,
      "step": 593
    },
    {
      "epoch": 0.75,
      "learning_rate": 4.462929475587704e-05,
      "loss": 0.274,
      "step": 594
    },
    {
      "epoch": 0.75,
      "learning_rate": 4.462025316455696e-05,
      "loss": 0.2876,
      "step": 595
    },
    {
      "epoch": 0.75,
      "learning_rate": 4.461121157323689e-05,
      "loss": 0.155,
      "step": 596
    },
    {
      "epoch": 0.76,
      "learning_rate": 4.460216998191682e-05,
      "loss": 0.4391,
      "step": 597
    },
    {
      "epoch": 0.76,
      "learning_rate": 4.459312839059674e-05,
      "loss": 0.2262,
      "step": 598
    },
    {
      "epoch": 0.76,
      "learning_rate": 4.4584086799276675e-05,
      "loss": 0.2016,
      "step": 599
    },
    {
      "epoch": 0.76,
      "learning_rate": 4.45750452079566e-05,
      "loss": 0.1592,
      "step": 600
    },
    {
      "epoch": 0.76,
      "learning_rate": 4.456600361663653e-05,
      "loss": 0.1994,
      "step": 601
    },
    {
      "epoch": 0.76,
      "learning_rate": 4.4556962025316456e-05,
      "loss": 0.3437,
      "step": 602
    },
    {
      "epoch": 0.76,
      "learning_rate": 4.454792043399639e-05,
      "loss": 0.2533,
      "step": 603
    },
    {
      "epoch": 0.76,
      "learning_rate": 4.453887884267631e-05,
      "loss": 0.2284,
      "step": 604
    },
    {
      "epoch": 0.77,
      "learning_rate": 4.4529837251356244e-05,
      "loss": 0.3612,
      "step": 605
    },
    {
      "epoch": 0.77,
      "learning_rate": 4.452079566003617e-05,
      "loss": 0.108,
      "step": 606
    },
    {
      "epoch": 0.77,
      "learning_rate": 4.4511754068716094e-05,
      "loss": 0.1745,
      "step": 607
    },
    {
      "epoch": 0.77,
      "learning_rate": 4.4502712477396025e-05,
      "loss": 0.1752,
      "step": 608
    },
    {
      "epoch": 0.77,
      "learning_rate": 4.449367088607595e-05,
      "loss": 0.1533,
      "step": 609
    },
    {
      "epoch": 0.77,
      "learning_rate": 4.448462929475588e-05,
      "loss": 0.1529,
      "step": 610
    },
    {
      "epoch": 0.77,
      "learning_rate": 4.447558770343581e-05,
      "loss": 0.2128,
      "step": 611
    },
    {
      "epoch": 0.77,
      "learning_rate": 4.446654611211574e-05,
      "loss": 0.221,
      "step": 612
    },
    {
      "epoch": 0.78,
      "learning_rate": 4.445750452079566e-05,
      "loss": 0.2575,
      "step": 613
    },
    {
      "epoch": 0.78,
      "learning_rate": 4.444846292947559e-05,
      "loss": 0.1163,
      "step": 614
    },
    {
      "epoch": 0.78,
      "learning_rate": 4.443942133815552e-05,
      "loss": 0.4603,
      "step": 615
    },
    {
      "epoch": 0.78,
      "learning_rate": 4.4430379746835445e-05,
      "loss": 0.1295,
      "step": 616
    },
    {
      "epoch": 0.78,
      "learning_rate": 4.4421338155515376e-05,
      "loss": 0.1654,
      "step": 617
    },
    {
      "epoch": 0.78,
      "learning_rate": 4.44122965641953e-05,
      "loss": 0.1874,
      "step": 618
    },
    {
      "epoch": 0.78,
      "learning_rate": 4.440325497287523e-05,
      "loss": 0.3087,
      "step": 619
    },
    {
      "epoch": 0.78,
      "learning_rate": 4.439421338155516e-05,
      "loss": 0.1847,
      "step": 620
    },
    {
      "epoch": 0.79,
      "learning_rate": 4.438517179023509e-05,
      "loss": 0.3249,
      "step": 621
    },
    {
      "epoch": 0.79,
      "learning_rate": 4.4376130198915014e-05,
      "loss": 0.2153,
      "step": 622
    },
    {
      "epoch": 0.79,
      "learning_rate": 4.436708860759494e-05,
      "loss": 0.1132,
      "step": 623
    },
    {
      "epoch": 0.79,
      "learning_rate": 4.435804701627487e-05,
      "loss": 0.4502,
      "step": 624
    },
    {
      "epoch": 0.79,
      "learning_rate": 4.4349005424954795e-05,
      "loss": 0.2886,
      "step": 625
    },
    {
      "epoch": 0.79,
      "learning_rate": 4.433996383363472e-05,
      "loss": 0.2546,
      "step": 626
    },
    {
      "epoch": 0.79,
      "learning_rate": 4.433092224231465e-05,
      "loss": 0.2995,
      "step": 627
    },
    {
      "epoch": 0.79,
      "learning_rate": 4.432188065099458e-05,
      "loss": 0.4892,
      "step": 628
    },
    {
      "epoch": 0.8,
      "learning_rate": 4.43128390596745e-05,
      "loss": 0.3243,
      "step": 629
    },
    {
      "epoch": 0.8,
      "learning_rate": 4.430379746835443e-05,
      "loss": 0.4532,
      "step": 630
    },
    {
      "epoch": 0.8,
      "learning_rate": 4.429475587703436e-05,
      "loss": 0.4199,
      "step": 631
    },
    {
      "epoch": 0.8,
      "learning_rate": 4.428571428571428e-05,
      "loss": 0.4124,
      "step": 632
    },
    {
      "epoch": 0.8,
      "learning_rate": 4.4276672694394214e-05,
      "loss": 0.324,
      "step": 633
    },
    {
      "epoch": 0.8,
      "learning_rate": 4.426763110307414e-05,
      "loss": 0.2941,
      "step": 634
    },
    {
      "epoch": 0.8,
      "learning_rate": 4.425858951175407e-05,
      "loss": 0.1331,
      "step": 635
    },
    {
      "epoch": 0.8,
      "learning_rate": 4.4249547920433996e-05,
      "loss": 0.1241,
      "step": 636
    },
    {
      "epoch": 0.81,
      "learning_rate": 4.424050632911393e-05,
      "loss": 0.1925,
      "step": 637
    },
    {
      "epoch": 0.81,
      "learning_rate": 4.423146473779385e-05,
      "loss": 0.2255,
      "step": 638
    },
    {
      "epoch": 0.81,
      "learning_rate": 4.422242314647378e-05,
      "loss": 0.3252,
      "step": 639
    },
    {
      "epoch": 0.81,
      "learning_rate": 4.421338155515371e-05,
      "loss": 0.2329,
      "step": 640
    },
    {
      "epoch": 0.81,
      "learning_rate": 4.4204339963833634e-05,
      "loss": 0.4144,
      "step": 641
    },
    {
      "epoch": 0.81,
      "learning_rate": 4.4195298372513565e-05,
      "loss": 0.1793,
      "step": 642
    },
    {
      "epoch": 0.81,
      "learning_rate": 4.418625678119349e-05,
      "loss": 0.0925,
      "step": 643
    },
    {
      "epoch": 0.81,
      "learning_rate": 4.417721518987342e-05,
      "loss": 0.2235,
      "step": 644
    },
    {
      "epoch": 0.82,
      "learning_rate": 4.4168173598553347e-05,
      "loss": 0.2594,
      "step": 645
    },
    {
      "epoch": 0.82,
      "learning_rate": 4.415913200723327e-05,
      "loss": 0.2454,
      "step": 646
    },
    {
      "epoch": 0.82,
      "learning_rate": 4.41500904159132e-05,
      "loss": 0.195,
      "step": 647
    },
    {
      "epoch": 0.82,
      "learning_rate": 4.414104882459313e-05,
      "loss": 0.2949,
      "step": 648
    },
    {
      "epoch": 0.82,
      "learning_rate": 4.413200723327306e-05,
      "loss": 0.292,
      "step": 649
    },
    {
      "epoch": 0.82,
      "learning_rate": 4.4122965641952984e-05,
      "loss": 0.2638,
      "step": 650
    },
    {
      "epoch": 0.82,
      "learning_rate": 4.4113924050632916e-05,
      "loss": 0.1049,
      "step": 651
    },
    {
      "epoch": 0.82,
      "learning_rate": 4.410488245931284e-05,
      "loss": 0.1919,
      "step": 652
    },
    {
      "epoch": 0.83,
      "learning_rate": 4.409584086799277e-05,
      "loss": 0.5058,
      "step": 653
    },
    {
      "epoch": 0.83,
      "learning_rate": 4.40867992766727e-05,
      "loss": 0.2299,
      "step": 654
    },
    {
      "epoch": 0.83,
      "learning_rate": 4.407775768535262e-05,
      "loss": 0.2351,
      "step": 655
    },
    {
      "epoch": 0.83,
      "learning_rate": 4.4068716094032554e-05,
      "loss": 0.2867,
      "step": 656
    },
    {
      "epoch": 0.83,
      "learning_rate": 4.405967450271248e-05,
      "loss": 0.4811,
      "step": 657
    },
    {
      "epoch": 0.83,
      "learning_rate": 4.405063291139241e-05,
      "loss": 0.1708,
      "step": 658
    },
    {
      "epoch": 0.83,
      "learning_rate": 4.4041591320072335e-05,
      "loss": 0.4568,
      "step": 659
    },
    {
      "epoch": 0.83,
      "learning_rate": 4.403254972875227e-05,
      "loss": 0.1995,
      "step": 660
    },
    {
      "epoch": 0.84,
      "learning_rate": 4.402350813743219e-05,
      "loss": 0.2553,
      "step": 661
    },
    {
      "epoch": 0.84,
      "learning_rate": 4.4014466546112116e-05,
      "loss": 0.4273,
      "step": 662
    },
    {
      "epoch": 0.84,
      "learning_rate": 4.400542495479205e-05,
      "loss": 0.3735,
      "step": 663
    },
    {
      "epoch": 0.84,
      "learning_rate": 4.399638336347197e-05,
      "loss": 0.2976,
      "step": 664
    },
    {
      "epoch": 0.84,
      "learning_rate": 4.3987341772151904e-05,
      "loss": 0.2869,
      "step": 665
    },
    {
      "epoch": 0.84,
      "learning_rate": 4.397830018083183e-05,
      "loss": 0.2563,
      "step": 666
    },
    {
      "epoch": 0.84,
      "learning_rate": 4.396925858951176e-05,
      "loss": 0.4143,
      "step": 667
    },
    {
      "epoch": 0.85,
      "learning_rate": 4.3960216998191686e-05,
      "loss": 0.2088,
      "step": 668
    },
    {
      "epoch": 0.85,
      "learning_rate": 4.395117540687162e-05,
      "loss": 0.2659,
      "step": 669
    },
    {
      "epoch": 0.85,
      "learning_rate": 4.394213381555154e-05,
      "loss": 0.5003,
      "step": 670
    },
    {
      "epoch": 0.85,
      "learning_rate": 4.393309222423147e-05,
      "loss": 0.1828,
      "step": 671
    },
    {
      "epoch": 0.85,
      "learning_rate": 4.39240506329114e-05,
      "loss": 0.2031,
      "step": 672
    },
    {
      "epoch": 0.85,
      "learning_rate": 4.3915009041591324e-05,
      "loss": 0.2337,
      "step": 673
    },
    {
      "epoch": 0.85,
      "learning_rate": 4.390596745027125e-05,
      "loss": 0.2456,
      "step": 674
    },
    {
      "epoch": 0.85,
      "learning_rate": 4.389692585895118e-05,
      "loss": 0.2997,
      "step": 675
    },
    {
      "epoch": 0.86,
      "learning_rate": 4.3887884267631105e-05,
      "loss": 0.2725,
      "step": 676
    },
    {
      "epoch": 0.86,
      "learning_rate": 4.387884267631103e-05,
      "loss": 0.2955,
      "step": 677
    },
    {
      "epoch": 0.86,
      "learning_rate": 4.3869801084990955e-05,
      "loss": 0.4639,
      "step": 678
    },
    {
      "epoch": 0.86,
      "learning_rate": 4.3860759493670886e-05,
      "loss": 0.2497,
      "step": 679
    },
    {
      "epoch": 0.86,
      "learning_rate": 4.385171790235081e-05,
      "loss": 0.3361,
      "step": 680
    },
    {
      "epoch": 0.86,
      "learning_rate": 4.384267631103074e-05,
      "loss": 0.2263,
      "step": 681
    },
    {
      "epoch": 0.86,
      "learning_rate": 4.383363471971067e-05,
      "loss": 0.1604,
      "step": 682
    },
    {
      "epoch": 0.86,
      "learning_rate": 4.38245931283906e-05,
      "loss": 0.3259,
      "step": 683
    },
    {
      "epoch": 0.87,
      "learning_rate": 4.3815551537070524e-05,
      "loss": 0.2348,
      "step": 684
    },
    {
      "epoch": 0.87,
      "learning_rate": 4.3806509945750456e-05,
      "loss": 0.1863,
      "step": 685
    },
    {
      "epoch": 0.87,
      "learning_rate": 4.379746835443038e-05,
      "loss": 0.1794,
      "step": 686
    },
    {
      "epoch": 0.87,
      "learning_rate": 4.3788426763110305e-05,
      "loss": 0.3618,
      "step": 687
    },
    {
      "epoch": 0.87,
      "learning_rate": 4.377938517179024e-05,
      "loss": 0.2422,
      "step": 688
    },
    {
      "epoch": 0.87,
      "learning_rate": 4.377034358047016e-05,
      "loss": 0.175,
      "step": 689
    },
    {
      "epoch": 0.87,
      "learning_rate": 4.3761301989150093e-05,
      "loss": 0.1752,
      "step": 690
    },
    {
      "epoch": 0.87,
      "learning_rate": 4.375226039783002e-05,
      "loss": 0.3296,
      "step": 691
    },
    {
      "epoch": 0.88,
      "learning_rate": 4.374321880650995e-05,
      "loss": 0.1845,
      "step": 692
    },
    {
      "epoch": 0.88,
      "learning_rate": 4.3734177215189875e-05,
      "loss": 0.344,
      "step": 693
    },
    {
      "epoch": 0.88,
      "learning_rate": 4.37251356238698e-05,
      "loss": 0.4797,
      "step": 694
    },
    {
      "epoch": 0.88,
      "learning_rate": 4.371609403254973e-05,
      "loss": 0.2312,
      "step": 695
    },
    {
      "epoch": 0.88,
      "learning_rate": 4.3707052441229656e-05,
      "loss": 0.1549,
      "step": 696
    },
    {
      "epoch": 0.88,
      "learning_rate": 4.369801084990959e-05,
      "loss": 0.0936,
      "step": 697
    },
    {
      "epoch": 0.88,
      "learning_rate": 4.368896925858951e-05,
      "loss": 0.1239,
      "step": 698
    },
    {
      "epoch": 0.88,
      "learning_rate": 4.3679927667269444e-05,
      "loss": 0.37,
      "step": 699
    },
    {
      "epoch": 0.89,
      "learning_rate": 4.367088607594937e-05,
      "loss": 0.0856,
      "step": 700
    },
    {
      "epoch": 0.89,
      "learning_rate": 4.36618444846293e-05,
      "loss": 0.1958,
      "step": 701
    },
    {
      "epoch": 0.89,
      "learning_rate": 4.3652802893309226e-05,
      "loss": 0.2569,
      "step": 702
    },
    {
      "epoch": 0.89,
      "learning_rate": 4.364376130198915e-05,
      "loss": 0.2215,
      "step": 703
    },
    {
      "epoch": 0.89,
      "learning_rate": 4.363471971066908e-05,
      "loss": 0.299,
      "step": 704
    },
    {
      "epoch": 0.89,
      "learning_rate": 4.362567811934901e-05,
      "loss": 0.1618,
      "step": 705
    },
    {
      "epoch": 0.89,
      "learning_rate": 4.361663652802894e-05,
      "loss": 0.1665,
      "step": 706
    },
    {
      "epoch": 0.89,
      "learning_rate": 4.360759493670886e-05,
      "loss": 0.1467,
      "step": 707
    },
    {
      "epoch": 0.9,
      "learning_rate": 4.3598553345388795e-05,
      "loss": 0.2205,
      "step": 708
    },
    {
      "epoch": 0.9,
      "learning_rate": 4.358951175406872e-05,
      "loss": 0.3356,
      "step": 709
    },
    {
      "epoch": 0.9,
      "learning_rate": 4.3580470162748645e-05,
      "loss": 0.201,
      "step": 710
    },
    {
      "epoch": 0.9,
      "learning_rate": 4.3571428571428576e-05,
      "loss": 0.2209,
      "step": 711
    },
    {
      "epoch": 0.9,
      "learning_rate": 4.35623869801085e-05,
      "loss": 0.1662,
      "step": 712
    },
    {
      "epoch": 0.9,
      "learning_rate": 4.355334538878843e-05,
      "loss": 0.1702,
      "step": 713
    },
    {
      "epoch": 0.9,
      "learning_rate": 4.354430379746836e-05,
      "loss": 0.2065,
      "step": 714
    },
    {
      "epoch": 0.9,
      "learning_rate": 4.353526220614829e-05,
      "loss": 0.2544,
      "step": 715
    },
    {
      "epoch": 0.91,
      "learning_rate": 4.3526220614828214e-05,
      "loss": 0.5203,
      "step": 716
    },
    {
      "epoch": 0.91,
      "learning_rate": 4.351717902350814e-05,
      "loss": 0.2547,
      "step": 717
    },
    {
      "epoch": 0.91,
      "learning_rate": 4.350813743218807e-05,
      "loss": 0.2614,
      "step": 718
    },
    {
      "epoch": 0.91,
      "learning_rate": 4.3499095840867995e-05,
      "loss": 0.3286,
      "step": 719
    },
    {
      "epoch": 0.91,
      "learning_rate": 4.349005424954792e-05,
      "loss": 0.0897,
      "step": 720
    },
    {
      "epoch": 0.91,
      "learning_rate": 4.348101265822785e-05,
      "loss": 0.1598,
      "step": 721
    },
    {
      "epoch": 0.91,
      "learning_rate": 4.347197106690778e-05,
      "loss": 0.2404,
      "step": 722
    },
    {
      "epoch": 0.91,
      "learning_rate": 4.34629294755877e-05,
      "loss": 0.0738,
      "step": 723
    },
    {
      "epoch": 0.92,
      "learning_rate": 4.345388788426763e-05,
      "loss": 0.5413,
      "step": 724
    },
    {
      "epoch": 0.92,
      "learning_rate": 4.344484629294756e-05,
      "loss": 0.2249,
      "step": 725
    },
    {
      "epoch": 0.92,
      "learning_rate": 4.343580470162748e-05,
      "loss": 0.3262,
      "step": 726
    },
    {
      "epoch": 0.92,
      "learning_rate": 4.3426763110307415e-05,
      "loss": 0.251,
      "step": 727
    },
    {
      "epoch": 0.92,
      "learning_rate": 4.341772151898734e-05,
      "loss": 0.492,
      "step": 728
    },
    {
      "epoch": 0.92,
      "learning_rate": 4.340867992766727e-05,
      "loss": 0.3625,
      "step": 729
    },
    {
      "epoch": 0.92,
      "learning_rate": 4.3399638336347196e-05,
      "loss": 0.064,
      "step": 730
    },
    {
      "epoch": 0.92,
      "learning_rate": 4.339059674502713e-05,
      "loss": 0.1696,
      "step": 731
    },
    {
      "epoch": 0.93,
      "learning_rate": 4.338155515370705e-05,
      "loss": 0.4462,
      "step": 732
    },
    {
      "epoch": 0.93,
      "learning_rate": 4.3372513562386984e-05,
      "loss": 0.1324,
      "step": 733
    },
    {
      "epoch": 0.93,
      "learning_rate": 4.336347197106691e-05,
      "loss": 0.2058,
      "step": 734
    },
    {
      "epoch": 0.93,
      "learning_rate": 4.3354430379746834e-05,
      "loss": 0.1149,
      "step": 735
    },
    {
      "epoch": 0.93,
      "learning_rate": 4.3345388788426765e-05,
      "loss": 0.3095,
      "step": 736
    },
    {
      "epoch": 0.93,
      "learning_rate": 4.333634719710669e-05,
      "loss": 0.0836,
      "step": 737
    },
    {
      "epoch": 0.93,
      "learning_rate": 4.332730560578662e-05,
      "loss": 0.3698,
      "step": 738
    },
    {
      "epoch": 0.93,
      "learning_rate": 4.331826401446655e-05,
      "loss": 0.2411,
      "step": 739
    },
    {
      "epoch": 0.94,
      "learning_rate": 4.330922242314648e-05,
      "loss": 0.3263,
      "step": 740
    },
    {
      "epoch": 0.94,
      "learning_rate": 4.33001808318264e-05,
      "loss": 0.1447,
      "step": 741
    },
    {
      "epoch": 0.94,
      "learning_rate": 4.329113924050633e-05,
      "loss": 0.2534,
      "step": 742
    },
    {
      "epoch": 0.94,
      "learning_rate": 4.328209764918626e-05,
      "loss": 0.4093,
      "step": 743
    },
    {
      "epoch": 0.94,
      "learning_rate": 4.3273056057866184e-05,
      "loss": 0.1519,
      "step": 744
    },
    {
      "epoch": 0.94,
      "learning_rate": 4.3264014466546116e-05,
      "loss": 0.2456,
      "step": 745
    },
    {
      "epoch": 0.94,
      "learning_rate": 4.325497287522604e-05,
      "loss": 0.5027,
      "step": 746
    },
    {
      "epoch": 0.94,
      "learning_rate": 4.324593128390597e-05,
      "loss": 0.14,
      "step": 747
    },
    {
      "epoch": 0.95,
      "learning_rate": 4.32368896925859e-05,
      "loss": 0.1531,
      "step": 748
    },
    {
      "epoch": 0.95,
      "learning_rate": 4.322784810126583e-05,
      "loss": 0.2386,
      "step": 749
    },
    {
      "epoch": 0.95,
      "learning_rate": 4.3218806509945754e-05,
      "loss": 0.2177,
      "step": 750
    },
    {
      "epoch": 0.95,
      "eval_accuracy": 0.9031531531531531,
      "eval_loss": 0.2327551394701004,
      "eval_runtime": 45.1987,
      "eval_samples_per_second": 29.47,
      "eval_steps_per_second": 3.695,
      "step": 750
    },
    {
      "epoch": 0.95,
      "learning_rate": 4.320976491862568e-05,
      "loss": 0.1909,
      "step": 751
    },
    {
      "epoch": 0.95,
      "learning_rate": 4.320072332730561e-05,
      "loss": 0.455,
      "step": 752
    },
    {
      "epoch": 0.95,
      "learning_rate": 4.3191681735985535e-05,
      "loss": 0.1262,
      "step": 753
    },
    {
      "epoch": 0.95,
      "learning_rate": 4.318264014466547e-05,
      "loss": 0.3197,
      "step": 754
    },
    {
      "epoch": 0.96,
      "learning_rate": 4.317359855334539e-05,
      "loss": 0.5704,
      "step": 755
    },
    {
      "epoch": 0.96,
      "learning_rate": 4.316455696202532e-05,
      "loss": 0.1544,
      "step": 756
    },
    {
      "epoch": 0.96,
      "learning_rate": 4.315551537070525e-05,
      "loss": 0.0597,
      "step": 757
    },
    {
      "epoch": 0.96,
      "learning_rate": 4.314647377938517e-05,
      "loss": 0.2083,
      "step": 758
    },
    {
      "epoch": 0.96,
      "learning_rate": 4.3137432188065105e-05,
      "loss": 0.2634,
      "step": 759
    },
    {
      "epoch": 0.96,
      "learning_rate": 4.312839059674503e-05,
      "loss": 0.2442,
      "step": 760
    },
    {
      "epoch": 0.96,
      "learning_rate": 4.311934900542496e-05,
      "loss": 0.3648,
      "step": 761
    },
    {
      "epoch": 0.96,
      "learning_rate": 4.3110307414104886e-05,
      "loss": 0.092,
      "step": 762
    },
    {
      "epoch": 0.97,
      "learning_rate": 4.310126582278482e-05,
      "loss": 0.4937,
      "step": 763
    },
    {
      "epoch": 0.97,
      "learning_rate": 4.309222423146474e-05,
      "loss": 0.2661,
      "step": 764
    },
    {
      "epoch": 0.97,
      "learning_rate": 4.308318264014467e-05,
      "loss": 0.3191,
      "step": 765
    },
    {
      "epoch": 0.97,
      "learning_rate": 4.30741410488246e-05,
      "loss": 0.1557,
      "step": 766
    },
    {
      "epoch": 0.97,
      "learning_rate": 4.3065099457504524e-05,
      "loss": 0.1828,
      "step": 767
    },
    {
      "epoch": 0.97,
      "learning_rate": 4.305605786618445e-05,
      "loss": 0.2967,
      "step": 768
    },
    {
      "epoch": 0.97,
      "learning_rate": 4.3047016274864373e-05,
      "loss": 0.1593,
      "step": 769
    },
    {
      "epoch": 0.97,
      "learning_rate": 4.3037974683544305e-05,
      "loss": 0.3102,
      "step": 770
    },
    {
      "epoch": 0.98,
      "learning_rate": 4.302893309222423e-05,
      "loss": 0.2767,
      "step": 771
    },
    {
      "epoch": 0.98,
      "learning_rate": 4.301989150090416e-05,
      "loss": 0.2421,
      "step": 772
    },
    {
      "epoch": 0.98,
      "learning_rate": 4.3010849909584086e-05,
      "loss": 0.4958,
      "step": 773
    },
    {
      "epoch": 0.98,
      "learning_rate": 4.300180831826401e-05,
      "loss": 0.2708,
      "step": 774
    },
    {
      "epoch": 0.98,
      "learning_rate": 4.299276672694394e-05,
      "loss": 0.2857,
      "step": 775
    },
    {
      "epoch": 0.98,
      "learning_rate": 4.298372513562387e-05,
      "loss": 0.1976,
      "step": 776
    },
    {
      "epoch": 0.98,
      "learning_rate": 4.29746835443038e-05,
      "loss": 0.2424,
      "step": 777
    },
    {
      "epoch": 0.98,
      "learning_rate": 4.2965641952983724e-05,
      "loss": 0.2332,
      "step": 778
    },
    {
      "epoch": 0.99,
      "learning_rate": 4.2956600361663656e-05,
      "loss": 0.0802,
      "step": 779
    },
    {
      "epoch": 0.99,
      "learning_rate": 4.294755877034358e-05,
      "loss": 0.1683,
      "step": 780
    },
    {
      "epoch": 0.99,
      "learning_rate": 4.2938517179023506e-05,
      "loss": 0.2065,
      "step": 781
    },
    {
      "epoch": 0.99,
      "learning_rate": 4.292947558770344e-05,
      "loss": 0.3103,
      "step": 782
    },
    {
      "epoch": 0.99,
      "learning_rate": 4.292043399638336e-05,
      "loss": 0.3683,
      "step": 783
    },
    {
      "epoch": 0.99,
      "learning_rate": 4.2911392405063294e-05,
      "loss": 0.1448,
      "step": 784
    },
    {
      "epoch": 0.99,
      "learning_rate": 4.290235081374322e-05,
      "loss": 0.1636,
      "step": 785
    },
    {
      "epoch": 0.99,
      "learning_rate": 4.289330922242315e-05,
      "loss": 0.1982,
      "step": 786
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.2884267631103075e-05,
      "loss": 0.3732,
      "step": 787
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.2875226039783007e-05,
      "loss": 0.2044,
      "step": 788
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.286618444846293e-05,
      "loss": 0.3319,
      "step": 789
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.2857142857142856e-05,
      "loss": 0.1502,
      "step": 790
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.284810126582279e-05,
      "loss": 0.4705,
      "step": 791
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.283905967450271e-05,
      "loss": 0.1385,
      "step": 792
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.2830018083182644e-05,
      "loss": 0.2278,
      "step": 793
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.282097649186257e-05,
      "loss": 0.1854,
      "step": 794
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.28119349005425e-05,
      "loss": 0.1094,
      "step": 795
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.2802893309222426e-05,
      "loss": 0.1445,
      "step": 796
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.279385171790235e-05,
      "loss": 0.079,
      "step": 797
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.278481012658228e-05,
      "loss": 0.2016,
      "step": 798
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.277576853526221e-05,
      "loss": 0.0827,
      "step": 799
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.276672694394214e-05,
      "loss": 0.2055,
      "step": 800
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.2757685352622064e-05,
      "loss": 0.2404,
      "step": 801
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.2748643761301995e-05,
      "loss": 0.062,
      "step": 802
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.273960216998192e-05,
      "loss": 0.3371,
      "step": 803
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.273056057866185e-05,
      "loss": 0.1801,
      "step": 804
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.2721518987341776e-05,
      "loss": 0.1904,
      "step": 805
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.27124773960217e-05,
      "loss": 0.2054,
      "step": 806
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.270343580470163e-05,
      "loss": 0.2435,
      "step": 807
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.269439421338156e-05,
      "loss": 0.1129,
      "step": 808
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.268535262206149e-05,
      "loss": 0.0666,
      "step": 809
    },
    {
      "epoch": 1.03,
      "learning_rate": 4.2676311030741414e-05,
      "loss": 0.2796,
      "step": 810
    },
    {
      "epoch": 1.03,
      "learning_rate": 4.266726943942134e-05,
      "loss": 0.2492,
      "step": 811
    },
    {
      "epoch": 1.03,
      "learning_rate": 4.265822784810127e-05,
      "loss": 0.0603,
      "step": 812
    },
    {
      "epoch": 1.03,
      "learning_rate": 4.2649186256781196e-05,
      "loss": 0.3246,
      "step": 813
    },
    {
      "epoch": 1.03,
      "learning_rate": 4.264014466546112e-05,
      "loss": 0.079,
      "step": 814
    },
    {
      "epoch": 1.03,
      "learning_rate": 4.263110307414105e-05,
      "loss": 0.0715,
      "step": 815
    },
    {
      "epoch": 1.03,
      "learning_rate": 4.262206148282098e-05,
      "loss": 0.1671,
      "step": 816
    },
    {
      "epoch": 1.03,
      "learning_rate": 4.26130198915009e-05,
      "loss": 0.2492,
      "step": 817
    },
    {
      "epoch": 1.04,
      "learning_rate": 4.2603978300180833e-05,
      "loss": 0.1329,
      "step": 818
    },
    {
      "epoch": 1.04,
      "learning_rate": 4.259493670886076e-05,
      "loss": 0.1324,
      "step": 819
    },
    {
      "epoch": 1.04,
      "learning_rate": 4.258589511754069e-05,
      "loss": 0.1789,
      "step": 820
    },
    {
      "epoch": 1.04,
      "learning_rate": 4.2576853526220615e-05,
      "loss": 0.1904,
      "step": 821
    },
    {
      "epoch": 1.04,
      "learning_rate": 4.256781193490054e-05,
      "loss": 0.3668,
      "step": 822
    },
    {
      "epoch": 1.04,
      "learning_rate": 4.255877034358047e-05,
      "loss": 0.2867,
      "step": 823
    },
    {
      "epoch": 1.04,
      "learning_rate": 4.2549728752260396e-05,
      "loss": 0.1227,
      "step": 824
    },
    {
      "epoch": 1.04,
      "learning_rate": 4.254068716094033e-05,
      "loss": 0.1046,
      "step": 825
    },
    {
      "epoch": 1.05,
      "learning_rate": 4.253164556962025e-05,
      "loss": 0.2148,
      "step": 826
    },
    {
      "epoch": 1.05,
      "learning_rate": 4.2522603978300184e-05,
      "loss": 0.2815,
      "step": 827
    },
    {
      "epoch": 1.05,
      "learning_rate": 4.251356238698011e-05,
      "loss": 0.111,
      "step": 828
    },
    {
      "epoch": 1.05,
      "learning_rate": 4.2504520795660034e-05,
      "loss": 0.1473,
      "step": 829
    },
    {
      "epoch": 1.05,
      "learning_rate": 4.2495479204339965e-05,
      "loss": 0.0532,
      "step": 830
    },
    {
      "epoch": 1.05,
      "learning_rate": 4.248643761301989e-05,
      "loss": 0.4346,
      "step": 831
    },
    {
      "epoch": 1.05,
      "learning_rate": 4.247739602169982e-05,
      "loss": 0.1394,
      "step": 832
    },
    {
      "epoch": 1.05,
      "learning_rate": 4.246835443037975e-05,
      "loss": 0.044,
      "step": 833
    },
    {
      "epoch": 1.06,
      "learning_rate": 4.245931283905968e-05,
      "loss": 0.1207,
      "step": 834
    },
    {
      "epoch": 1.06,
      "learning_rate": 4.24502712477396e-05,
      "loss": 0.2527,
      "step": 835
    },
    {
      "epoch": 1.06,
      "learning_rate": 4.2441229656419535e-05,
      "loss": 0.4699,
      "step": 836
    },
    {
      "epoch": 1.06,
      "learning_rate": 4.243218806509946e-05,
      "loss": 0.3796,
      "step": 837
    },
    {
      "epoch": 1.06,
      "learning_rate": 4.2423146473779385e-05,
      "loss": 0.1434,
      "step": 838
    },
    {
      "epoch": 1.06,
      "learning_rate": 4.2414104882459316e-05,
      "loss": 0.2668,
      "step": 839
    },
    {
      "epoch": 1.06,
      "learning_rate": 4.240506329113924e-05,
      "loss": 0.1069,
      "step": 840
    },
    {
      "epoch": 1.06,
      "learning_rate": 4.239602169981917e-05,
      "loss": 0.1223,
      "step": 841
    },
    {
      "epoch": 1.07,
      "learning_rate": 4.23869801084991e-05,
      "loss": 0.2366,
      "step": 842
    },
    {
      "epoch": 1.07,
      "learning_rate": 4.237793851717903e-05,
      "loss": 0.3152,
      "step": 843
    },
    {
      "epoch": 1.07,
      "learning_rate": 4.2368896925858954e-05,
      "loss": 0.1955,
      "step": 844
    },
    {
      "epoch": 1.07,
      "learning_rate": 4.235985533453888e-05,
      "loss": 0.4097,
      "step": 845
    },
    {
      "epoch": 1.07,
      "learning_rate": 4.235081374321881e-05,
      "loss": 0.3158,
      "step": 846
    },
    {
      "epoch": 1.07,
      "learning_rate": 4.2341772151898735e-05,
      "loss": 0.2201,
      "step": 847
    },
    {
      "epoch": 1.07,
      "learning_rate": 4.233273056057867e-05,
      "loss": 0.1791,
      "step": 848
    },
    {
      "epoch": 1.07,
      "learning_rate": 4.232368896925859e-05,
      "loss": 0.248,
      "step": 849
    },
    {
      "epoch": 1.08,
      "learning_rate": 4.2314647377938523e-05,
      "loss": 0.2252,
      "step": 850
    },
    {
      "epoch": 1.08,
      "learning_rate": 4.230560578661845e-05,
      "loss": 0.2018,
      "step": 851
    },
    {
      "epoch": 1.08,
      "learning_rate": 4.229656419529838e-05,
      "loss": 0.1871,
      "step": 852
    },
    {
      "epoch": 1.08,
      "learning_rate": 4.2287522603978305e-05,
      "loss": 0.2294,
      "step": 853
    },
    {
      "epoch": 1.08,
      "learning_rate": 4.227848101265823e-05,
      "loss": 0.1982,
      "step": 854
    },
    {
      "epoch": 1.08,
      "learning_rate": 4.226943942133816e-05,
      "loss": 0.2457,
      "step": 855
    },
    {
      "epoch": 1.08,
      "learning_rate": 4.2260397830018086e-05,
      "loss": 0.2285,
      "step": 856
    },
    {
      "epoch": 1.08,
      "learning_rate": 4.225135623869802e-05,
      "loss": 0.2392,
      "step": 857
    },
    {
      "epoch": 1.09,
      "learning_rate": 4.224231464737794e-05,
      "loss": 0.5754,
      "step": 858
    },
    {
      "epoch": 1.09,
      "learning_rate": 4.223327305605787e-05,
      "loss": 0.0596,
      "step": 859
    },
    {
      "epoch": 1.09,
      "learning_rate": 4.22242314647378e-05,
      "loss": 0.2319,
      "step": 860
    },
    {
      "epoch": 1.09,
      "learning_rate": 4.2215189873417724e-05,
      "loss": 0.233,
      "step": 861
    },
    {
      "epoch": 1.09,
      "learning_rate": 4.220614828209765e-05,
      "loss": 0.2692,
      "step": 862
    },
    {
      "epoch": 1.09,
      "learning_rate": 4.2197106690777574e-05,
      "loss": 0.4576,
      "step": 863
    },
    {
      "epoch": 1.09,
      "learning_rate": 4.2188065099457505e-05,
      "loss": 0.105,
      "step": 864
    },
    {
      "epoch": 1.09,
      "learning_rate": 4.217902350813743e-05,
      "loss": 0.0569,
      "step": 865
    },
    {
      "epoch": 1.1,
      "learning_rate": 4.216998191681736e-05,
      "loss": 0.2446,
      "step": 866
    },
    {
      "epoch": 1.1,
      "learning_rate": 4.2160940325497287e-05,
      "loss": 0.3832,
      "step": 867
    },
    {
      "epoch": 1.1,
      "learning_rate": 4.215189873417722e-05,
      "loss": 0.065,
      "step": 868
    },
    {
      "epoch": 1.1,
      "learning_rate": 4.214285714285714e-05,
      "loss": 0.0457,
      "step": 869
    },
    {
      "epoch": 1.1,
      "learning_rate": 4.213381555153707e-05,
      "loss": 0.1882,
      "step": 870
    },
    {
      "epoch": 1.1,
      "learning_rate": 4.2124773960217e-05,
      "loss": 0.1702,
      "step": 871
    },
    {
      "epoch": 1.1,
      "learning_rate": 4.2115732368896924e-05,
      "loss": 0.0816,
      "step": 872
    },
    {
      "epoch": 1.1,
      "learning_rate": 4.2106690777576856e-05,
      "loss": 0.4113,
      "step": 873
    },
    {
      "epoch": 1.11,
      "learning_rate": 4.209764918625678e-05,
      "loss": 0.2148,
      "step": 874
    },
    {
      "epoch": 1.11,
      "learning_rate": 4.208860759493671e-05,
      "loss": 0.2211,
      "step": 875
    },
    {
      "epoch": 1.11,
      "learning_rate": 4.207956600361664e-05,
      "loss": 0.3201,
      "step": 876
    },
    {
      "epoch": 1.11,
      "learning_rate": 4.207052441229656e-05,
      "loss": 0.2486,
      "step": 877
    },
    {
      "epoch": 1.11,
      "learning_rate": 4.2061482820976494e-05,
      "loss": 0.1323,
      "step": 878
    },
    {
      "epoch": 1.11,
      "learning_rate": 4.205244122965642e-05,
      "loss": 0.3962,
      "step": 879
    },
    {
      "epoch": 1.11,
      "learning_rate": 4.204339963833635e-05,
      "loss": 0.1218,
      "step": 880
    },
    {
      "epoch": 1.12,
      "learning_rate": 4.2034358047016275e-05,
      "loss": 0.1739,
      "step": 881
    },
    {
      "epoch": 1.12,
      "learning_rate": 4.202531645569621e-05,
      "loss": 0.1923,
      "step": 882
    },
    {
      "epoch": 1.12,
      "learning_rate": 4.201627486437613e-05,
      "loss": 0.0811,
      "step": 883
    },
    {
      "epoch": 1.12,
      "learning_rate": 4.200723327305606e-05,
      "loss": 0.2233,
      "step": 884
    },
    {
      "epoch": 1.12,
      "learning_rate": 4.199819168173599e-05,
      "loss": 0.1152,
      "step": 885
    },
    {
      "epoch": 1.12,
      "learning_rate": 4.198915009041591e-05,
      "loss": 0.2581,
      "step": 886
    },
    {
      "epoch": 1.12,
      "learning_rate": 4.1980108499095845e-05,
      "loss": 0.1621,
      "step": 887
    },
    {
      "epoch": 1.12,
      "learning_rate": 4.197106690777577e-05,
      "loss": 0.2735,
      "step": 888
    },
    {
      "epoch": 1.13,
      "learning_rate": 4.19620253164557e-05,
      "loss": 0.1722,
      "step": 889
    },
    {
      "epoch": 1.13,
      "learning_rate": 4.1952983725135626e-05,
      "loss": 0.0839,
      "step": 890
    },
    {
      "epoch": 1.13,
      "learning_rate": 4.194394213381556e-05,
      "loss": 0.2955,
      "step": 891
    },
    {
      "epoch": 1.13,
      "learning_rate": 4.193490054249548e-05,
      "loss": 0.1344,
      "step": 892
    },
    {
      "epoch": 1.13,
      "learning_rate": 4.192585895117541e-05,
      "loss": 0.1286,
      "step": 893
    },
    {
      "epoch": 1.13,
      "learning_rate": 4.191681735985534e-05,
      "loss": 0.2257,
      "step": 894
    },
    {
      "epoch": 1.13,
      "learning_rate": 4.1907775768535264e-05,
      "loss": 0.2003,
      "step": 895
    },
    {
      "epoch": 1.13,
      "learning_rate": 4.1898734177215195e-05,
      "loss": 0.1644,
      "step": 896
    },
    {
      "epoch": 1.14,
      "learning_rate": 4.188969258589512e-05,
      "loss": 0.4682,
      "step": 897
    },
    {
      "epoch": 1.14,
      "learning_rate": 4.188065099457505e-05,
      "loss": 0.2466,
      "step": 898
    },
    {
      "epoch": 1.14,
      "learning_rate": 4.187160940325498e-05,
      "loss": 0.2265,
      "step": 899
    },
    {
      "epoch": 1.14,
      "learning_rate": 4.186256781193491e-05,
      "loss": 0.233,
      "step": 900
    },
    {
      "epoch": 1.14,
      "learning_rate": 4.185352622061483e-05,
      "loss": 0.3018,
      "step": 901
    },
    {
      "epoch": 1.14,
      "learning_rate": 4.184448462929476e-05,
      "loss": 0.1216,
      "step": 902
    },
    {
      "epoch": 1.14,
      "learning_rate": 4.183544303797469e-05,
      "loss": 0.317,
      "step": 903
    },
    {
      "epoch": 1.14,
      "learning_rate": 4.1826401446654614e-05,
      "loss": 0.2597,
      "step": 904
    },
    {
      "epoch": 1.15,
      "learning_rate": 4.181735985533454e-05,
      "loss": 0.101,
      "step": 905
    },
    {
      "epoch": 1.15,
      "learning_rate": 4.180831826401447e-05,
      "loss": 0.2191,
      "step": 906
    },
    {
      "epoch": 1.15,
      "learning_rate": 4.1799276672694396e-05,
      "loss": 0.0849,
      "step": 907
    },
    {
      "epoch": 1.15,
      "learning_rate": 4.179023508137432e-05,
      "loss": 0.1189,
      "step": 908
    },
    {
      "epoch": 1.15,
      "learning_rate": 4.178119349005425e-05,
      "loss": 0.1554,
      "step": 909
    },
    {
      "epoch": 1.15,
      "learning_rate": 4.177215189873418e-05,
      "loss": 0.4277,
      "step": 910
    },
    {
      "epoch": 1.15,
      "learning_rate": 4.17631103074141e-05,
      "loss": 0.1144,
      "step": 911
    },
    {
      "epoch": 1.15,
      "learning_rate": 4.1754068716094034e-05,
      "loss": 0.0807,
      "step": 912
    },
    {
      "epoch": 1.16,
      "learning_rate": 4.174502712477396e-05,
      "loss": 0.1841,
      "step": 913
    },
    {
      "epoch": 1.16,
      "learning_rate": 4.173598553345389e-05,
      "loss": 0.2367,
      "step": 914
    },
    {
      "epoch": 1.16,
      "learning_rate": 4.1726943942133815e-05,
      "loss": 0.2213,
      "step": 915
    },
    {
      "epoch": 1.16,
      "learning_rate": 4.1717902350813746e-05,
      "loss": 0.1367,
      "step": 916
    },
    {
      "epoch": 1.16,
      "learning_rate": 4.170886075949367e-05,
      "loss": 0.1538,
      "step": 917
    },
    {
      "epoch": 1.16,
      "learning_rate": 4.1699819168173596e-05,
      "loss": 0.1966,
      "step": 918
    },
    {
      "epoch": 1.16,
      "learning_rate": 4.169077757685353e-05,
      "loss": 0.1789,
      "step": 919
    },
    {
      "epoch": 1.16,
      "learning_rate": 4.168173598553345e-05,
      "loss": 0.2618,
      "step": 920
    },
    {
      "epoch": 1.17,
      "learning_rate": 4.1672694394213384e-05,
      "loss": 0.1803,
      "step": 921
    },
    {
      "epoch": 1.17,
      "learning_rate": 4.166365280289331e-05,
      "loss": 0.0744,
      "step": 922
    },
    {
      "epoch": 1.17,
      "learning_rate": 4.165461121157324e-05,
      "loss": 0.2672,
      "step": 923
    },
    {
      "epoch": 1.17,
      "learning_rate": 4.1645569620253166e-05,
      "loss": 0.1379,
      "step": 924
    },
    {
      "epoch": 1.17,
      "learning_rate": 4.163652802893309e-05,
      "loss": 0.1714,
      "step": 925
    },
    {
      "epoch": 1.17,
      "learning_rate": 4.162748643761302e-05,
      "loss": 0.1278,
      "step": 926
    },
    {
      "epoch": 1.17,
      "learning_rate": 4.161844484629295e-05,
      "loss": 0.1752,
      "step": 927
    },
    {
      "epoch": 1.17,
      "learning_rate": 4.160940325497288e-05,
      "loss": 0.1845,
      "step": 928
    },
    {
      "epoch": 1.18,
      "learning_rate": 4.1600361663652803e-05,
      "loss": 0.2042,
      "step": 929
    },
    {
      "epoch": 1.18,
      "learning_rate": 4.1591320072332735e-05,
      "loss": 0.1032,
      "step": 930
    },
    {
      "epoch": 1.18,
      "learning_rate": 4.158227848101266e-05,
      "loss": 0.2195,
      "step": 931
    },
    {
      "epoch": 1.18,
      "learning_rate": 4.157323688969259e-05,
      "loss": 0.1066,
      "step": 932
    },
    {
      "epoch": 1.18,
      "learning_rate": 4.1564195298372516e-05,
      "loss": 0.1352,
      "step": 933
    },
    {
      "epoch": 1.18,
      "learning_rate": 4.155515370705244e-05,
      "loss": 0.4267,
      "step": 934
    },
    {
      "epoch": 1.18,
      "learning_rate": 4.154611211573237e-05,
      "loss": 0.1578,
      "step": 935
    },
    {
      "epoch": 1.18,
      "learning_rate": 4.15370705244123e-05,
      "loss": 0.1025,
      "step": 936
    },
    {
      "epoch": 1.19,
      "learning_rate": 4.152802893309223e-05,
      "loss": 0.407,
      "step": 937
    },
    {
      "epoch": 1.19,
      "learning_rate": 4.1518987341772154e-05,
      "loss": 0.1485,
      "step": 938
    },
    {
      "epoch": 1.19,
      "learning_rate": 4.1509945750452086e-05,
      "loss": 0.1758,
      "step": 939
    },
    {
      "epoch": 1.19,
      "learning_rate": 4.150090415913201e-05,
      "loss": 0.1715,
      "step": 940
    },
    {
      "epoch": 1.19,
      "learning_rate": 4.1491862567811936e-05,
      "loss": 0.1786,
      "step": 941
    },
    {
      "epoch": 1.19,
      "learning_rate": 4.148282097649187e-05,
      "loss": 0.197,
      "step": 942
    },
    {
      "epoch": 1.19,
      "learning_rate": 4.147377938517179e-05,
      "loss": 0.0606,
      "step": 943
    },
    {
      "epoch": 1.19,
      "learning_rate": 4.1464737793851724e-05,
      "loss": 0.3023,
      "step": 944
    },
    {
      "epoch": 1.2,
      "learning_rate": 4.145569620253165e-05,
      "loss": 0.0744,
      "step": 945
    },
    {
      "epoch": 1.2,
      "learning_rate": 4.144665461121158e-05,
      "loss": 0.2886,
      "step": 946
    },
    {
      "epoch": 1.2,
      "learning_rate": 4.1437613019891505e-05,
      "loss": 0.2524,
      "step": 947
    },
    {
      "epoch": 1.2,
      "learning_rate": 4.1428571428571437e-05,
      "loss": 0.169,
      "step": 948
    },
    {
      "epoch": 1.2,
      "learning_rate": 4.141952983725136e-05,
      "loss": 0.192,
      "step": 949
    },
    {
      "epoch": 1.2,
      "learning_rate": 4.1410488245931286e-05,
      "loss": 0.1802,
      "step": 950
    },
    {
      "epoch": 1.2,
      "learning_rate": 4.140144665461122e-05,
      "loss": 0.2464,
      "step": 951
    },
    {
      "epoch": 1.2,
      "learning_rate": 4.139240506329114e-05,
      "loss": 0.3391,
      "step": 952
    },
    {
      "epoch": 1.21,
      "learning_rate": 4.138336347197107e-05,
      "loss": 0.254,
      "step": 953
    },
    {
      "epoch": 1.21,
      "learning_rate": 4.137432188065099e-05,
      "loss": 0.1207,
      "step": 954
    },
    {
      "epoch": 1.21,
      "learning_rate": 4.1365280289330924e-05,
      "loss": 0.06,
      "step": 955
    },
    {
      "epoch": 1.21,
      "learning_rate": 4.135623869801085e-05,
      "loss": 0.2689,
      "step": 956
    },
    {
      "epoch": 1.21,
      "learning_rate": 4.1347197106690774e-05,
      "loss": 0.2608,
      "step": 957
    },
    {
      "epoch": 1.21,
      "learning_rate": 4.1338155515370705e-05,
      "loss": 0.1784,
      "step": 958
    },
    {
      "epoch": 1.21,
      "learning_rate": 4.132911392405063e-05,
      "loss": 0.1033,
      "step": 959
    },
    {
      "epoch": 1.22,
      "learning_rate": 4.132007233273056e-05,
      "loss": 0.1539,
      "step": 960
    },
    {
      "epoch": 1.22,
      "learning_rate": 4.131103074141049e-05,
      "loss": 0.1192,
      "step": 961
    },
    {
      "epoch": 1.22,
      "learning_rate": 4.130198915009042e-05,
      "loss": 0.1992,
      "step": 962
    },
    {
      "epoch": 1.22,
      "learning_rate": 4.129294755877034e-05,
      "loss": 0.3103,
      "step": 963
    },
    {
      "epoch": 1.22,
      "learning_rate": 4.1283905967450275e-05,
      "loss": 0.1006,
      "step": 964
    },
    {
      "epoch": 1.22,
      "learning_rate": 4.12748643761302e-05,
      "loss": 0.1404,
      "step": 965
    },
    {
      "epoch": 1.22,
      "learning_rate": 4.1265822784810125e-05,
      "loss": 0.0578,
      "step": 966
    },
    {
      "epoch": 1.22,
      "learning_rate": 4.1256781193490056e-05,
      "loss": 0.0719,
      "step": 967
    },
    {
      "epoch": 1.23,
      "learning_rate": 4.124773960216998e-05,
      "loss": 0.1445,
      "step": 968
    },
    {
      "epoch": 1.23,
      "learning_rate": 4.123869801084991e-05,
      "loss": 0.1978,
      "step": 969
    },
    {
      "epoch": 1.23,
      "learning_rate": 4.122965641952984e-05,
      "loss": 0.209,
      "step": 970
    },
    {
      "epoch": 1.23,
      "learning_rate": 4.122061482820977e-05,
      "loss": 0.0181,
      "step": 971
    },
    {
      "epoch": 1.23,
      "learning_rate": 4.1211573236889694e-05,
      "loss": 0.428,
      "step": 972
    },
    {
      "epoch": 1.23,
      "learning_rate": 4.120253164556962e-05,
      "loss": 0.333,
      "step": 973
    },
    {
      "epoch": 1.23,
      "learning_rate": 4.119349005424955e-05,
      "loss": 0.0454,
      "step": 974
    },
    {
      "epoch": 1.23,
      "learning_rate": 4.1184448462929475e-05,
      "loss": 0.1562,
      "step": 975
    },
    {
      "epoch": 1.24,
      "learning_rate": 4.117540687160941e-05,
      "loss": 0.0228,
      "step": 976
    },
    {
      "epoch": 1.24,
      "learning_rate": 4.116636528028933e-05,
      "loss": 0.1659,
      "step": 977
    },
    {
      "epoch": 1.24,
      "learning_rate": 4.115732368896926e-05,
      "loss": 0.2158,
      "step": 978
    },
    {
      "epoch": 1.24,
      "learning_rate": 4.114828209764919e-05,
      "loss": 0.1829,
      "step": 979
    },
    {
      "epoch": 1.24,
      "learning_rate": 4.113924050632912e-05,
      "loss": 0.1251,
      "step": 980
    },
    {
      "epoch": 1.24,
      "learning_rate": 4.1130198915009045e-05,
      "loss": 0.2313,
      "step": 981
    },
    {
      "epoch": 1.24,
      "learning_rate": 4.112115732368897e-05,
      "loss": 0.1602,
      "step": 982
    },
    {
      "epoch": 1.24,
      "learning_rate": 4.11121157323689e-05,
      "loss": 0.549,
      "step": 983
    },
    {
      "epoch": 1.25,
      "learning_rate": 4.1103074141048826e-05,
      "loss": 0.3298,
      "step": 984
    },
    {
      "epoch": 1.25,
      "learning_rate": 4.109403254972876e-05,
      "loss": 0.331,
      "step": 985
    },
    {
      "epoch": 1.25,
      "learning_rate": 4.108499095840868e-05,
      "loss": 0.0487,
      "step": 986
    },
    {
      "epoch": 1.25,
      "learning_rate": 4.1075949367088614e-05,
      "loss": 0.2433,
      "step": 987
    },
    {
      "epoch": 1.25,
      "learning_rate": 4.106690777576854e-05,
      "loss": 0.2585,
      "step": 988
    },
    {
      "epoch": 1.25,
      "learning_rate": 4.1057866184448464e-05,
      "loss": 0.1845,
      "step": 989
    },
    {
      "epoch": 1.25,
      "learning_rate": 4.1048824593128395e-05,
      "loss": 0.3355,
      "step": 990
    },
    {
      "epoch": 1.25,
      "learning_rate": 4.103978300180832e-05,
      "loss": 0.3988,
      "step": 991
    },
    {
      "epoch": 1.26,
      "learning_rate": 4.103074141048825e-05,
      "loss": 0.1505,
      "step": 992
    },
    {
      "epoch": 1.26,
      "learning_rate": 4.102169981916818e-05,
      "loss": 0.1732,
      "step": 993
    },
    {
      "epoch": 1.26,
      "learning_rate": 4.101265822784811e-05,
      "loss": 0.1406,
      "step": 994
    },
    {
      "epoch": 1.26,
      "learning_rate": 4.100361663652803e-05,
      "loss": 0.2364,
      "step": 995
    },
    {
      "epoch": 1.26,
      "learning_rate": 4.099457504520796e-05,
      "loss": 0.2579,
      "step": 996
    },
    {
      "epoch": 1.26,
      "learning_rate": 4.098553345388789e-05,
      "loss": 0.1032,
      "step": 997
    },
    {
      "epoch": 1.26,
      "learning_rate": 4.0976491862567815e-05,
      "loss": 0.1278,
      "step": 998
    },
    {
      "epoch": 1.26,
      "learning_rate": 4.096745027124774e-05,
      "loss": 0.1367,
      "step": 999
    },
    {
      "epoch": 1.27,
      "learning_rate": 4.095840867992767e-05,
      "loss": 0.3212,
      "step": 1000
    },
    {
      "epoch": 1.27,
      "eval_accuracy": 0.9024024024024024,
      "eval_loss": 0.2397383451461792,
      "eval_runtime": 45.1086,
      "eval_samples_per_second": 29.529,
      "eval_steps_per_second": 3.702,
      "step": 1000
    },
    {
      "epoch": 1.27,
      "learning_rate": 4.0949367088607596e-05,
      "loss": 0.2819,
      "step": 1001
    },
    {
      "epoch": 1.27,
      "learning_rate": 4.094032549728752e-05,
      "loss": 0.1003,
      "step": 1002
    },
    {
      "epoch": 1.27,
      "learning_rate": 4.093128390596745e-05,
      "loss": 0.1212,
      "step": 1003
    },
    {
      "epoch": 1.27,
      "learning_rate": 4.092224231464738e-05,
      "loss": 0.4578,
      "step": 1004
    },
    {
      "epoch": 1.27,
      "learning_rate": 4.09132007233273e-05,
      "loss": 0.1889,
      "step": 1005
    },
    {
      "epoch": 1.27,
      "learning_rate": 4.0904159132007234e-05,
      "loss": 0.193,
      "step": 1006
    },
    {
      "epoch": 1.27,
      "learning_rate": 4.089511754068716e-05,
      "loss": 0.2463,
      "step": 1007
    },
    {
      "epoch": 1.28,
      "learning_rate": 4.088607594936709e-05,
      "loss": 0.1436,
      "step": 1008
    },
    {
      "epoch": 1.28,
      "learning_rate": 4.0877034358047015e-05,
      "loss": 0.115,
      "step": 1009
    },
    {
      "epoch": 1.28,
      "learning_rate": 4.086799276672695e-05,
      "loss": 0.3216,
      "step": 1010
    },
    {
      "epoch": 1.28,
      "learning_rate": 4.085895117540687e-05,
      "loss": 0.2286,
      "step": 1011
    },
    {
      "epoch": 1.28,
      "learning_rate": 4.0849909584086796e-05,
      "loss": 0.1608,
      "step": 1012
    },
    {
      "epoch": 1.28,
      "learning_rate": 4.084086799276673e-05,
      "loss": 0.3679,
      "step": 1013
    },
    {
      "epoch": 1.28,
      "learning_rate": 4.083182640144665e-05,
      "loss": 0.2512,
      "step": 1014
    },
    {
      "epoch": 1.28,
      "learning_rate": 4.0822784810126584e-05,
      "loss": 0.0244,
      "step": 1015
    },
    {
      "epoch": 1.29,
      "learning_rate": 4.081374321880651e-05,
      "loss": 0.2194,
      "step": 1016
    },
    {
      "epoch": 1.29,
      "learning_rate": 4.080470162748644e-05,
      "loss": 0.3781,
      "step": 1017
    },
    {
      "epoch": 1.29,
      "learning_rate": 4.0795660036166366e-05,
      "loss": 0.3032,
      "step": 1018
    },
    {
      "epoch": 1.29,
      "learning_rate": 4.07866184448463e-05,
      "loss": 0.1947,
      "step": 1019
    },
    {
      "epoch": 1.29,
      "learning_rate": 4.077757685352622e-05,
      "loss": 0.2532,
      "step": 1020
    },
    {
      "epoch": 1.29,
      "learning_rate": 4.076853526220615e-05,
      "loss": 0.2131,
      "step": 1021
    },
    {
      "epoch": 1.29,
      "learning_rate": 4.075949367088608e-05,
      "loss": 0.1786,
      "step": 1022
    },
    {
      "epoch": 1.29,
      "learning_rate": 4.0750452079566004e-05,
      "loss": 0.149,
      "step": 1023
    },
    {
      "epoch": 1.3,
      "learning_rate": 4.0741410488245935e-05,
      "loss": 0.2742,
      "step": 1024
    },
    {
      "epoch": 1.3,
      "learning_rate": 4.073236889692586e-05,
      "loss": 0.2905,
      "step": 1025
    },
    {
      "epoch": 1.3,
      "learning_rate": 4.072332730560579e-05,
      "loss": 0.2441,
      "step": 1026
    },
    {
      "epoch": 1.3,
      "learning_rate": 4.0714285714285717e-05,
      "loss": 0.4147,
      "step": 1027
    },
    {
      "epoch": 1.3,
      "learning_rate": 4.070524412296564e-05,
      "loss": 0.3969,
      "step": 1028
    },
    {
      "epoch": 1.3,
      "learning_rate": 4.069620253164557e-05,
      "loss": 0.1047,
      "step": 1029
    },
    {
      "epoch": 1.3,
      "learning_rate": 4.06871609403255e-05,
      "loss": 0.14,
      "step": 1030
    },
    {
      "epoch": 1.3,
      "learning_rate": 4.067811934900543e-05,
      "loss": 0.399,
      "step": 1031
    },
    {
      "epoch": 1.31,
      "learning_rate": 4.0669077757685354e-05,
      "loss": 0.1259,
      "step": 1032
    },
    {
      "epoch": 1.31,
      "learning_rate": 4.0660036166365286e-05,
      "loss": 0.1263,
      "step": 1033
    },
    {
      "epoch": 1.31,
      "learning_rate": 4.065099457504521e-05,
      "loss": 0.1555,
      "step": 1034
    },
    {
      "epoch": 1.31,
      "learning_rate": 4.064195298372514e-05,
      "loss": 0.1874,
      "step": 1035
    },
    {
      "epoch": 1.31,
      "learning_rate": 4.063291139240507e-05,
      "loss": 0.2445,
      "step": 1036
    },
    {
      "epoch": 1.31,
      "learning_rate": 4.062386980108499e-05,
      "loss": 0.1686,
      "step": 1037
    },
    {
      "epoch": 1.31,
      "learning_rate": 4.0614828209764924e-05,
      "loss": 0.1311,
      "step": 1038
    },
    {
      "epoch": 1.31,
      "learning_rate": 4.060578661844485e-05,
      "loss": 0.184,
      "step": 1039
    },
    {
      "epoch": 1.32,
      "learning_rate": 4.059674502712478e-05,
      "loss": 0.2275,
      "step": 1040
    },
    {
      "epoch": 1.32,
      "learning_rate": 4.0587703435804705e-05,
      "loss": 0.0878,
      "step": 1041
    },
    {
      "epoch": 1.32,
      "learning_rate": 4.057866184448464e-05,
      "loss": 0.333,
      "step": 1042
    },
    {
      "epoch": 1.32,
      "learning_rate": 4.056962025316456e-05,
      "loss": 0.2163,
      "step": 1043
    },
    {
      "epoch": 1.32,
      "learning_rate": 4.0560578661844486e-05,
      "loss": 0.0913,
      "step": 1044
    },
    {
      "epoch": 1.32,
      "learning_rate": 4.055153707052441e-05,
      "loss": 0.0549,
      "step": 1045
    },
    {
      "epoch": 1.32,
      "learning_rate": 4.054249547920434e-05,
      "loss": 0.219,
      "step": 1046
    },
    {
      "epoch": 1.33,
      "learning_rate": 4.053345388788427e-05,
      "loss": 0.1375,
      "step": 1047
    },
    {
      "epoch": 1.33,
      "learning_rate": 4.052441229656419e-05,
      "loss": 0.1048,
      "step": 1048
    },
    {
      "epoch": 1.33,
      "learning_rate": 4.0515370705244124e-05,
      "loss": 0.1882,
      "step": 1049
    },
    {
      "epoch": 1.33,
      "learning_rate": 4.050632911392405e-05,
      "loss": 0.1251,
      "step": 1050
    },
    {
      "epoch": 1.33,
      "learning_rate": 4.049728752260398e-05,
      "loss": 0.0934,
      "step": 1051
    },
    {
      "epoch": 1.33,
      "learning_rate": 4.0488245931283906e-05,
      "loss": 0.1958,
      "step": 1052
    },
    {
      "epoch": 1.33,
      "learning_rate": 4.047920433996383e-05,
      "loss": 0.1532,
      "step": 1053
    },
    {
      "epoch": 1.33,
      "learning_rate": 4.047016274864376e-05,
      "loss": 0.1402,
      "step": 1054
    },
    {
      "epoch": 1.34,
      "learning_rate": 4.046112115732369e-05,
      "loss": 0.44,
      "step": 1055
    },
    {
      "epoch": 1.34,
      "learning_rate": 4.045207956600362e-05,
      "loss": 0.0611,
      "step": 1056
    },
    {
      "epoch": 1.34,
      "learning_rate": 4.044303797468354e-05,
      "loss": 0.1356,
      "step": 1057
    },
    {
      "epoch": 1.34,
      "learning_rate": 4.0433996383363475e-05,
      "loss": 0.2533,
      "step": 1058
    },
    {
      "epoch": 1.34,
      "learning_rate": 4.04249547920434e-05,
      "loss": 0.2198,
      "step": 1059
    },
    {
      "epoch": 1.34,
      "learning_rate": 4.0415913200723325e-05,
      "loss": 0.2936,
      "step": 1060
    },
    {
      "epoch": 1.34,
      "learning_rate": 4.0406871609403256e-05,
      "loss": 0.174,
      "step": 1061
    },
    {
      "epoch": 1.34,
      "learning_rate": 4.039783001808318e-05,
      "loss": 0.0743,
      "step": 1062
    },
    {
      "epoch": 1.35,
      "learning_rate": 4.038878842676311e-05,
      "loss": 0.2121,
      "step": 1063
    },
    {
      "epoch": 1.35,
      "learning_rate": 4.037974683544304e-05,
      "loss": 0.1223,
      "step": 1064
    },
    {
      "epoch": 1.35,
      "learning_rate": 4.037070524412297e-05,
      "loss": 0.0761,
      "step": 1065
    },
    {
      "epoch": 1.35,
      "learning_rate": 4.0361663652802894e-05,
      "loss": 0.1145,
      "step": 1066
    },
    {
      "epoch": 1.35,
      "learning_rate": 4.0352622061482826e-05,
      "loss": 0.159,
      "step": 1067
    },
    {
      "epoch": 1.35,
      "learning_rate": 4.034358047016275e-05,
      "loss": 0.2513,
      "step": 1068
    },
    {
      "epoch": 1.35,
      "learning_rate": 4.0334538878842675e-05,
      "loss": 0.2331,
      "step": 1069
    },
    {
      "epoch": 1.35,
      "learning_rate": 4.032549728752261e-05,
      "loss": 0.1896,
      "step": 1070
    },
    {
      "epoch": 1.36,
      "learning_rate": 4.031645569620253e-05,
      "loss": 0.2575,
      "step": 1071
    },
    {
      "epoch": 1.36,
      "learning_rate": 4.0307414104882464e-05,
      "loss": 0.1616,
      "step": 1072
    },
    {
      "epoch": 1.36,
      "learning_rate": 4.029837251356239e-05,
      "loss": 0.1463,
      "step": 1073
    },
    {
      "epoch": 1.36,
      "learning_rate": 4.028933092224232e-05,
      "loss": 0.0814,
      "step": 1074
    },
    {
      "epoch": 1.36,
      "learning_rate": 4.0280289330922245e-05,
      "loss": 0.1972,
      "step": 1075
    },
    {
      "epoch": 1.36,
      "learning_rate": 4.027124773960217e-05,
      "loss": 0.0951,
      "step": 1076
    },
    {
      "epoch": 1.36,
      "learning_rate": 4.02622061482821e-05,
      "loss": 0.2064,
      "step": 1077
    },
    {
      "epoch": 1.36,
      "learning_rate": 4.0253164556962026e-05,
      "loss": 0.1086,
      "step": 1078
    },
    {
      "epoch": 1.37,
      "learning_rate": 4.024412296564196e-05,
      "loss": 0.2437,
      "step": 1079
    },
    {
      "epoch": 1.37,
      "learning_rate": 4.023508137432188e-05,
      "loss": 0.0751,
      "step": 1080
    },
    {
      "epoch": 1.37,
      "learning_rate": 4.0226039783001814e-05,
      "loss": 0.2362,
      "step": 1081
    },
    {
      "epoch": 1.37,
      "learning_rate": 4.021699819168174e-05,
      "loss": 0.3132,
      "step": 1082
    },
    {
      "epoch": 1.37,
      "learning_rate": 4.020795660036167e-05,
      "loss": 0.4062,
      "step": 1083
    },
    {
      "epoch": 1.37,
      "learning_rate": 4.0198915009041596e-05,
      "loss": 0.213,
      "step": 1084
    },
    {
      "epoch": 1.37,
      "learning_rate": 4.018987341772152e-05,
      "loss": 0.2646,
      "step": 1085
    },
    {
      "epoch": 1.37,
      "learning_rate": 4.018083182640145e-05,
      "loss": 0.2509,
      "step": 1086
    },
    {
      "epoch": 1.38,
      "learning_rate": 4.017179023508138e-05,
      "loss": 0.1007,
      "step": 1087
    },
    {
      "epoch": 1.38,
      "learning_rate": 4.016274864376131e-05,
      "loss": 0.0897,
      "step": 1088
    },
    {
      "epoch": 1.38,
      "learning_rate": 4.0153707052441233e-05,
      "loss": 0.1415,
      "step": 1089
    },
    {
      "epoch": 1.38,
      "learning_rate": 4.014466546112116e-05,
      "loss": 0.2023,
      "step": 1090
    },
    {
      "epoch": 1.38,
      "learning_rate": 4.013562386980109e-05,
      "loss": 0.1331,
      "step": 1091
    },
    {
      "epoch": 1.38,
      "learning_rate": 4.0126582278481015e-05,
      "loss": 0.2408,
      "step": 1092
    },
    {
      "epoch": 1.38,
      "learning_rate": 4.011754068716094e-05,
      "loss": 0.0999,
      "step": 1093
    },
    {
      "epoch": 1.38,
      "learning_rate": 4.010849909584087e-05,
      "loss": 0.1651,
      "step": 1094
    },
    {
      "epoch": 1.39,
      "learning_rate": 4.0099457504520796e-05,
      "loss": 0.1916,
      "step": 1095
    },
    {
      "epoch": 1.39,
      "learning_rate": 4.009041591320072e-05,
      "loss": 0.1082,
      "step": 1096
    },
    {
      "epoch": 1.39,
      "learning_rate": 4.008137432188065e-05,
      "loss": 0.1784,
      "step": 1097
    },
    {
      "epoch": 1.39,
      "learning_rate": 4.007233273056058e-05,
      "loss": 0.0786,
      "step": 1098
    },
    {
      "epoch": 1.39,
      "learning_rate": 4.006329113924051e-05,
      "loss": 0.1738,
      "step": 1099
    },
    {
      "epoch": 1.39,
      "learning_rate": 4.0054249547920434e-05,
      "loss": 0.2994,
      "step": 1100
    },
    {
      "epoch": 1.39,
      "learning_rate": 4.004520795660036e-05,
      "loss": 0.0899,
      "step": 1101
    },
    {
      "epoch": 1.39,
      "learning_rate": 4.003616636528029e-05,
      "loss": 0.1583,
      "step": 1102
    },
    {
      "epoch": 1.4,
      "learning_rate": 4.0027124773960215e-05,
      "loss": 0.2682,
      "step": 1103
    },
    {
      "epoch": 1.4,
      "learning_rate": 4.001808318264015e-05,
      "loss": 0.1437,
      "step": 1104
    },
    {
      "epoch": 1.4,
      "learning_rate": 4.000904159132007e-05,
      "loss": 0.3549,
      "step": 1105
    },
    {
      "epoch": 1.4,
      "learning_rate": 4e-05,
      "loss": 0.1804,
      "step": 1106
    },
    {
      "epoch": 1.4,
      "learning_rate": 3.999095840867993e-05,
      "loss": 0.0959,
      "step": 1107
    },
    {
      "epoch": 1.4,
      "learning_rate": 3.998191681735985e-05,
      "loss": 0.2961,
      "step": 1108
    },
    {
      "epoch": 1.4,
      "learning_rate": 3.9972875226039785e-05,
      "loss": 0.1615,
      "step": 1109
    },
    {
      "epoch": 1.4,
      "learning_rate": 3.996383363471971e-05,
      "loss": 0.0422,
      "step": 1110
    },
    {
      "epoch": 1.41,
      "learning_rate": 3.995479204339964e-05,
      "loss": 0.0742,
      "step": 1111
    },
    {
      "epoch": 1.41,
      "learning_rate": 3.9945750452079566e-05,
      "loss": 0.2517,
      "step": 1112
    },
    {
      "epoch": 1.41,
      "learning_rate": 3.99367088607595e-05,
      "loss": 0.0563,
      "step": 1113
    },
    {
      "epoch": 1.41,
      "learning_rate": 3.992766726943942e-05,
      "loss": 0.3838,
      "step": 1114
    },
    {
      "epoch": 1.41,
      "learning_rate": 3.9918625678119354e-05,
      "loss": 0.3113,
      "step": 1115
    },
    {
      "epoch": 1.41,
      "learning_rate": 3.990958408679928e-05,
      "loss": 0.1944,
      "step": 1116
    },
    {
      "epoch": 1.41,
      "learning_rate": 3.9900542495479204e-05,
      "loss": 0.3309,
      "step": 1117
    },
    {
      "epoch": 1.41,
      "learning_rate": 3.9891500904159135e-05,
      "loss": 0.3363,
      "step": 1118
    },
    {
      "epoch": 1.42,
      "learning_rate": 3.988245931283906e-05,
      "loss": 0.0849,
      "step": 1119
    },
    {
      "epoch": 1.42,
      "learning_rate": 3.987341772151899e-05,
      "loss": 0.5044,
      "step": 1120
    },
    {
      "epoch": 1.42,
      "learning_rate": 3.986437613019892e-05,
      "loss": 0.1629,
      "step": 1121
    },
    {
      "epoch": 1.42,
      "learning_rate": 3.985533453887885e-05,
      "loss": 0.2762,
      "step": 1122
    },
    {
      "epoch": 1.42,
      "learning_rate": 3.984629294755877e-05,
      "loss": 0.1543,
      "step": 1123
    },
    {
      "epoch": 1.42,
      "learning_rate": 3.98372513562387e-05,
      "loss": 0.0968,
      "step": 1124
    },
    {
      "epoch": 1.42,
      "learning_rate": 3.982820976491863e-05,
      "loss": 0.1912,
      "step": 1125
    },
    {
      "epoch": 1.43,
      "learning_rate": 3.9819168173598554e-05,
      "loss": 0.1964,
      "step": 1126
    },
    {
      "epoch": 1.43,
      "learning_rate": 3.9810126582278486e-05,
      "loss": 0.2125,
      "step": 1127
    },
    {
      "epoch": 1.43,
      "learning_rate": 3.980108499095841e-05,
      "loss": 0.1386,
      "step": 1128
    },
    {
      "epoch": 1.43,
      "learning_rate": 3.979204339963834e-05,
      "loss": 0.2153,
      "step": 1129
    },
    {
      "epoch": 1.43,
      "learning_rate": 3.978300180831827e-05,
      "loss": 0.2624,
      "step": 1130
    },
    {
      "epoch": 1.43,
      "learning_rate": 3.97739602169982e-05,
      "loss": 0.0652,
      "step": 1131
    },
    {
      "epoch": 1.43,
      "learning_rate": 3.9764918625678124e-05,
      "loss": 0.1016,
      "step": 1132
    },
    {
      "epoch": 1.43,
      "learning_rate": 3.975587703435805e-05,
      "loss": 0.077,
      "step": 1133
    },
    {
      "epoch": 1.44,
      "learning_rate": 3.974683544303798e-05,
      "loss": 0.081,
      "step": 1134
    },
    {
      "epoch": 1.44,
      "learning_rate": 3.9737793851717905e-05,
      "loss": 0.1236,
      "step": 1135
    },
    {
      "epoch": 1.44,
      "learning_rate": 3.972875226039784e-05,
      "loss": 0.0899,
      "step": 1136
    },
    {
      "epoch": 1.44,
      "learning_rate": 3.971971066907776e-05,
      "loss": 0.37,
      "step": 1137
    },
    {
      "epoch": 1.44,
      "learning_rate": 3.9710669077757687e-05,
      "loss": 0.3441,
      "step": 1138
    },
    {
      "epoch": 1.44,
      "learning_rate": 3.970162748643761e-05,
      "loss": 0.4805,
      "step": 1139
    },
    {
      "epoch": 1.44,
      "learning_rate": 3.969258589511754e-05,
      "loss": 0.3245,
      "step": 1140
    },
    {
      "epoch": 1.44,
      "learning_rate": 3.968354430379747e-05,
      "loss": 0.1494,
      "step": 1141
    },
    {
      "epoch": 1.45,
      "learning_rate": 3.967450271247739e-05,
      "loss": 0.2722,
      "step": 1142
    },
    {
      "epoch": 1.45,
      "learning_rate": 3.9665461121157324e-05,
      "loss": 0.1005,
      "step": 1143
    },
    {
      "epoch": 1.45,
      "learning_rate": 3.965641952983725e-05,
      "loss": 0.2117,
      "step": 1144
    },
    {
      "epoch": 1.45,
      "learning_rate": 3.964737793851718e-05,
      "loss": 0.1977,
      "step": 1145
    },
    {
      "epoch": 1.45,
      "learning_rate": 3.9638336347197106e-05,
      "loss": 0.2813,
      "step": 1146
    },
    {
      "epoch": 1.45,
      "learning_rate": 3.962929475587704e-05,
      "loss": 0.1179,
      "step": 1147
    },
    {
      "epoch": 1.45,
      "learning_rate": 3.962025316455696e-05,
      "loss": 0.0372,
      "step": 1148
    },
    {
      "epoch": 1.45,
      "learning_rate": 3.961121157323689e-05,
      "loss": 0.0897,
      "step": 1149
    },
    {
      "epoch": 1.46,
      "learning_rate": 3.960216998191682e-05,
      "loss": 0.1888,
      "step": 1150
    },
    {
      "epoch": 1.46,
      "learning_rate": 3.9593128390596743e-05,
      "loss": 0.3087,
      "step": 1151
    },
    {
      "epoch": 1.46,
      "learning_rate": 3.9584086799276675e-05,
      "loss": 0.1724,
      "step": 1152
    },
    {
      "epoch": 1.46,
      "learning_rate": 3.95750452079566e-05,
      "loss": 0.2786,
      "step": 1153
    },
    {
      "epoch": 1.46,
      "learning_rate": 3.956600361663653e-05,
      "loss": 0.2204,
      "step": 1154
    },
    {
      "epoch": 1.46,
      "learning_rate": 3.9556962025316456e-05,
      "loss": 0.2408,
      "step": 1155
    },
    {
      "epoch": 1.46,
      "learning_rate": 3.954792043399638e-05,
      "loss": 0.2734,
      "step": 1156
    },
    {
      "epoch": 1.46,
      "learning_rate": 3.953887884267631e-05,
      "loss": 0.3076,
      "step": 1157
    },
    {
      "epoch": 1.47,
      "learning_rate": 3.952983725135624e-05,
      "loss": 0.2134,
      "step": 1158
    },
    {
      "epoch": 1.47,
      "learning_rate": 3.952079566003617e-05,
      "loss": 0.2628,
      "step": 1159
    },
    {
      "epoch": 1.47,
      "learning_rate": 3.9511754068716094e-05,
      "loss": 0.3263,
      "step": 1160
    },
    {
      "epoch": 1.47,
      "learning_rate": 3.9502712477396026e-05,
      "loss": 0.087,
      "step": 1161
    },
    {
      "epoch": 1.47,
      "learning_rate": 3.949367088607595e-05,
      "loss": 0.2433,
      "step": 1162
    },
    {
      "epoch": 1.47,
      "learning_rate": 3.948462929475588e-05,
      "loss": 0.1948,
      "step": 1163
    },
    {
      "epoch": 1.47,
      "learning_rate": 3.947558770343581e-05,
      "loss": 0.1768,
      "step": 1164
    },
    {
      "epoch": 1.47,
      "learning_rate": 3.946654611211573e-05,
      "loss": 0.031,
      "step": 1165
    },
    {
      "epoch": 1.48,
      "learning_rate": 3.9457504520795664e-05,
      "loss": 0.154,
      "step": 1166
    },
    {
      "epoch": 1.48,
      "learning_rate": 3.944846292947559e-05,
      "loss": 0.1178,
      "step": 1167
    },
    {
      "epoch": 1.48,
      "learning_rate": 3.943942133815552e-05,
      "loss": 0.0522,
      "step": 1168
    },
    {
      "epoch": 1.48,
      "learning_rate": 3.9430379746835445e-05,
      "loss": 0.0621,
      "step": 1169
    },
    {
      "epoch": 1.48,
      "learning_rate": 3.942133815551538e-05,
      "loss": 0.4244,
      "step": 1170
    },
    {
      "epoch": 1.48,
      "learning_rate": 3.94122965641953e-05,
      "loss": 0.2072,
      "step": 1171
    },
    {
      "epoch": 1.48,
      "learning_rate": 3.9403254972875226e-05,
      "loss": 0.2399,
      "step": 1172
    },
    {
      "epoch": 1.48,
      "learning_rate": 3.939421338155516e-05,
      "loss": 0.0803,
      "step": 1173
    },
    {
      "epoch": 1.49,
      "learning_rate": 3.938517179023508e-05,
      "loss": 0.0399,
      "step": 1174
    },
    {
      "epoch": 1.49,
      "learning_rate": 3.9376130198915014e-05,
      "loss": 0.4306,
      "step": 1175
    },
    {
      "epoch": 1.49,
      "learning_rate": 3.936708860759494e-05,
      "loss": 0.1512,
      "step": 1176
    },
    {
      "epoch": 1.49,
      "learning_rate": 3.935804701627487e-05,
      "loss": 0.3579,
      "step": 1177
    },
    {
      "epoch": 1.49,
      "learning_rate": 3.9349005424954796e-05,
      "loss": 0.1334,
      "step": 1178
    },
    {
      "epoch": 1.49,
      "learning_rate": 3.933996383363473e-05,
      "loss": 0.2593,
      "step": 1179
    },
    {
      "epoch": 1.49,
      "learning_rate": 3.933092224231465e-05,
      "loss": 0.2921,
      "step": 1180
    },
    {
      "epoch": 1.49,
      "learning_rate": 3.932188065099458e-05,
      "loss": 0.1626,
      "step": 1181
    },
    {
      "epoch": 1.5,
      "learning_rate": 3.931283905967451e-05,
      "loss": 0.2798,
      "step": 1182
    },
    {
      "epoch": 1.5,
      "learning_rate": 3.9303797468354434e-05,
      "loss": 0.2977,
      "step": 1183
    },
    {
      "epoch": 1.5,
      "learning_rate": 3.929475587703436e-05,
      "loss": 0.1682,
      "step": 1184
    },
    {
      "epoch": 1.5,
      "learning_rate": 3.928571428571429e-05,
      "loss": 0.4678,
      "step": 1185
    },
    {
      "epoch": 1.5,
      "learning_rate": 3.9276672694394215e-05,
      "loss": 0.1273,
      "step": 1186
    },
    {
      "epoch": 1.5,
      "learning_rate": 3.926763110307414e-05,
      "loss": 0.1674,
      "step": 1187
    },
    {
      "epoch": 1.5,
      "learning_rate": 3.925858951175407e-05,
      "loss": 0.1438,
      "step": 1188
    },
    {
      "epoch": 1.5,
      "learning_rate": 3.9249547920433996e-05,
      "loss": 0.2457,
      "step": 1189
    },
    {
      "epoch": 1.51,
      "learning_rate": 3.924050632911392e-05,
      "loss": 0.1731,
      "step": 1190
    },
    {
      "epoch": 1.51,
      "learning_rate": 3.923146473779385e-05,
      "loss": 0.0994,
      "step": 1191
    },
    {
      "epoch": 1.51,
      "learning_rate": 3.922242314647378e-05,
      "loss": 0.1721,
      "step": 1192
    },
    {
      "epoch": 1.51,
      "learning_rate": 3.921338155515371e-05,
      "loss": 0.4863,
      "step": 1193
    },
    {
      "epoch": 1.51,
      "learning_rate": 3.9204339963833634e-05,
      "loss": 0.2582,
      "step": 1194
    },
    {
      "epoch": 1.51,
      "learning_rate": 3.9195298372513566e-05,
      "loss": 0.349,
      "step": 1195
    },
    {
      "epoch": 1.51,
      "learning_rate": 3.918625678119349e-05,
      "loss": 0.2071,
      "step": 1196
    },
    {
      "epoch": 1.51,
      "learning_rate": 3.9177215189873415e-05,
      "loss": 0.3539,
      "step": 1197
    },
    {
      "epoch": 1.52,
      "learning_rate": 3.916817359855335e-05,
      "loss": 0.2324,
      "step": 1198
    },
    {
      "epoch": 1.52,
      "learning_rate": 3.915913200723327e-05,
      "loss": 0.2998,
      "step": 1199
    },
    {
      "epoch": 1.52,
      "learning_rate": 3.9150090415913203e-05,
      "loss": 0.2416,
      "step": 1200
    },
    {
      "epoch": 1.52,
      "learning_rate": 3.914104882459313e-05,
      "loss": 0.2895,
      "step": 1201
    },
    {
      "epoch": 1.52,
      "learning_rate": 3.913200723327306e-05,
      "loss": 0.1353,
      "step": 1202
    },
    {
      "epoch": 1.52,
      "learning_rate": 3.9122965641952985e-05,
      "loss": 0.1094,
      "step": 1203
    },
    {
      "epoch": 1.52,
      "learning_rate": 3.911392405063291e-05,
      "loss": 0.1307,
      "step": 1204
    },
    {
      "epoch": 1.52,
      "learning_rate": 3.910488245931284e-05,
      "loss": 0.194,
      "step": 1205
    },
    {
      "epoch": 1.53,
      "learning_rate": 3.9095840867992766e-05,
      "loss": 0.0842,
      "step": 1206
    },
    {
      "epoch": 1.53,
      "learning_rate": 3.90867992766727e-05,
      "loss": 0.1919,
      "step": 1207
    },
    {
      "epoch": 1.53,
      "learning_rate": 3.907775768535262e-05,
      "loss": 0.2941,
      "step": 1208
    },
    {
      "epoch": 1.53,
      "learning_rate": 3.9068716094032554e-05,
      "loss": 0.1371,
      "step": 1209
    },
    {
      "epoch": 1.53,
      "learning_rate": 3.905967450271248e-05,
      "loss": 0.3866,
      "step": 1210
    },
    {
      "epoch": 1.53,
      "learning_rate": 3.905063291139241e-05,
      "loss": 0.3084,
      "step": 1211
    },
    {
      "epoch": 1.53,
      "learning_rate": 3.9041591320072335e-05,
      "loss": 0.08,
      "step": 1212
    },
    {
      "epoch": 1.54,
      "learning_rate": 3.903254972875226e-05,
      "loss": 0.0809,
      "step": 1213
    },
    {
      "epoch": 1.54,
      "learning_rate": 3.902350813743219e-05,
      "loss": 0.0375,
      "step": 1214
    },
    {
      "epoch": 1.54,
      "learning_rate": 3.901446654611212e-05,
      "loss": 0.1158,
      "step": 1215
    },
    {
      "epoch": 1.54,
      "learning_rate": 3.900542495479205e-05,
      "loss": 0.0977,
      "step": 1216
    },
    {
      "epoch": 1.54,
      "learning_rate": 3.899638336347197e-05,
      "loss": 0.1596,
      "step": 1217
    },
    {
      "epoch": 1.54,
      "learning_rate": 3.8987341772151905e-05,
      "loss": 0.1616,
      "step": 1218
    },
    {
      "epoch": 1.54,
      "learning_rate": 3.897830018083183e-05,
      "loss": 0.1601,
      "step": 1219
    },
    {
      "epoch": 1.54,
      "learning_rate": 3.8969258589511755e-05,
      "loss": 0.2659,
      "step": 1220
    },
    {
      "epoch": 1.55,
      "learning_rate": 3.8960216998191686e-05,
      "loss": 0.1681,
      "step": 1221
    },
    {
      "epoch": 1.55,
      "learning_rate": 3.895117540687161e-05,
      "loss": 0.3077,
      "step": 1222
    },
    {
      "epoch": 1.55,
      "learning_rate": 3.894213381555154e-05,
      "loss": 0.0904,
      "step": 1223
    },
    {
      "epoch": 1.55,
      "learning_rate": 3.893309222423147e-05,
      "loss": 0.0641,
      "step": 1224
    },
    {
      "epoch": 1.55,
      "learning_rate": 3.89240506329114e-05,
      "loss": 0.0949,
      "step": 1225
    },
    {
      "epoch": 1.55,
      "learning_rate": 3.8915009041591324e-05,
      "loss": 0.2934,
      "step": 1226
    },
    {
      "epoch": 1.55,
      "learning_rate": 3.8905967450271256e-05,
      "loss": 0.2355,
      "step": 1227
    },
    {
      "epoch": 1.55,
      "learning_rate": 3.889692585895118e-05,
      "loss": 0.0458,
      "step": 1228
    },
    {
      "epoch": 1.56,
      "learning_rate": 3.8887884267631105e-05,
      "loss": 0.1849,
      "step": 1229
    },
    {
      "epoch": 1.56,
      "learning_rate": 3.887884267631103e-05,
      "loss": 0.3154,
      "step": 1230
    },
    {
      "epoch": 1.56,
      "learning_rate": 3.886980108499096e-05,
      "loss": 0.3169,
      "step": 1231
    },
    {
      "epoch": 1.56,
      "learning_rate": 3.886075949367089e-05,
      "loss": 0.3535,
      "step": 1232
    },
    {
      "epoch": 1.56,
      "learning_rate": 3.885171790235081e-05,
      "loss": 0.2583,
      "step": 1233
    },
    {
      "epoch": 1.56,
      "learning_rate": 3.884267631103074e-05,
      "loss": 0.3767,
      "step": 1234
    },
    {
      "epoch": 1.56,
      "learning_rate": 3.883363471971067e-05,
      "loss": 0.048,
      "step": 1235
    },
    {
      "epoch": 1.56,
      "learning_rate": 3.882459312839059e-05,
      "loss": 0.1408,
      "step": 1236
    },
    {
      "epoch": 1.57,
      "learning_rate": 3.8815551537070525e-05,
      "loss": 0.1027,
      "step": 1237
    },
    {
      "epoch": 1.57,
      "learning_rate": 3.880650994575045e-05,
      "loss": 0.2109,
      "step": 1238
    },
    {
      "epoch": 1.57,
      "learning_rate": 3.879746835443038e-05,
      "loss": 0.1885,
      "step": 1239
    },
    {
      "epoch": 1.57,
      "learning_rate": 3.8788426763110306e-05,
      "loss": 0.1858,
      "step": 1240
    },
    {
      "epoch": 1.57,
      "learning_rate": 3.877938517179024e-05,
      "loss": 0.2072,
      "step": 1241
    },
    {
      "epoch": 1.57,
      "learning_rate": 3.877034358047016e-05,
      "loss": 0.2206,
      "step": 1242
    },
    {
      "epoch": 1.57,
      "learning_rate": 3.8761301989150094e-05,
      "loss": 0.3775,
      "step": 1243
    },
    {
      "epoch": 1.57,
      "learning_rate": 3.875226039783002e-05,
      "loss": 0.2638,
      "step": 1244
    },
    {
      "epoch": 1.58,
      "learning_rate": 3.8743218806509944e-05,
      "loss": 0.5664,
      "step": 1245
    },
    {
      "epoch": 1.58,
      "learning_rate": 3.8734177215189875e-05,
      "loss": 0.1153,
      "step": 1246
    },
    {
      "epoch": 1.58,
      "learning_rate": 3.87251356238698e-05,
      "loss": 0.1421,
      "step": 1247
    },
    {
      "epoch": 1.58,
      "learning_rate": 3.871609403254973e-05,
      "loss": 0.1025,
      "step": 1248
    },
    {
      "epoch": 1.58,
      "learning_rate": 3.8707052441229657e-05,
      "loss": 0.1451,
      "step": 1249
    },
    {
      "epoch": 1.58,
      "learning_rate": 3.869801084990959e-05,
      "loss": 0.4432,
      "step": 1250
    },
    {
      "epoch": 1.58,
      "eval_accuracy": 0.9054054054054054,
      "eval_loss": 0.23113574087619781,
      "eval_runtime": 45.1567,
      "eval_samples_per_second": 29.497,
      "eval_steps_per_second": 3.698,
      "step": 1250
    },
    {
      "epoch": 1.58,
      "learning_rate": 3.868896925858951e-05,
      "loss": 0.2196,
      "step": 1251
    },
    {
      "epoch": 1.58,
      "learning_rate": 3.867992766726944e-05,
      "loss": 0.194,
      "step": 1252
    },
    {
      "epoch": 1.59,
      "learning_rate": 3.867088607594937e-05,
      "loss": 0.144,
      "step": 1253
    },
    {
      "epoch": 1.59,
      "learning_rate": 3.8661844484629294e-05,
      "loss": 0.2175,
      "step": 1254
    },
    {
      "epoch": 1.59,
      "learning_rate": 3.8652802893309226e-05,
      "loss": 0.2454,
      "step": 1255
    },
    {
      "epoch": 1.59,
      "learning_rate": 3.864376130198915e-05,
      "loss": 0.3426,
      "step": 1256
    },
    {
      "epoch": 1.59,
      "learning_rate": 3.863471971066908e-05,
      "loss": 0.1213,
      "step": 1257
    },
    {
      "epoch": 1.59,
      "learning_rate": 3.862567811934901e-05,
      "loss": 0.2712,
      "step": 1258
    },
    {
      "epoch": 1.59,
      "learning_rate": 3.861663652802893e-05,
      "loss": 0.2387,
      "step": 1259
    },
    {
      "epoch": 1.59,
      "learning_rate": 3.8607594936708864e-05,
      "loss": 0.2572,
      "step": 1260
    },
    {
      "epoch": 1.6,
      "learning_rate": 3.859855334538879e-05,
      "loss": 0.2127,
      "step": 1261
    },
    {
      "epoch": 1.6,
      "learning_rate": 3.858951175406872e-05,
      "loss": 0.1699,
      "step": 1262
    },
    {
      "epoch": 1.6,
      "learning_rate": 3.8580470162748645e-05,
      "loss": 0.3462,
      "step": 1263
    },
    {
      "epoch": 1.6,
      "learning_rate": 3.857142857142858e-05,
      "loss": 0.2817,
      "step": 1264
    },
    {
      "epoch": 1.6,
      "learning_rate": 3.85623869801085e-05,
      "loss": 0.1913,
      "step": 1265
    },
    {
      "epoch": 1.6,
      "learning_rate": 3.855334538878843e-05,
      "loss": 0.0835,
      "step": 1266
    },
    {
      "epoch": 1.6,
      "learning_rate": 3.854430379746836e-05,
      "loss": 0.4658,
      "step": 1267
    },
    {
      "epoch": 1.6,
      "learning_rate": 3.853526220614828e-05,
      "loss": 0.2196,
      "step": 1268
    },
    {
      "epoch": 1.61,
      "learning_rate": 3.8526220614828215e-05,
      "loss": 0.2615,
      "step": 1269
    },
    {
      "epoch": 1.61,
      "learning_rate": 3.851717902350814e-05,
      "loss": 0.3934,
      "step": 1270
    },
    {
      "epoch": 1.61,
      "learning_rate": 3.850813743218807e-05,
      "loss": 0.169,
      "step": 1271
    },
    {
      "epoch": 1.61,
      "learning_rate": 3.8499095840867996e-05,
      "loss": 0.0742,
      "step": 1272
    },
    {
      "epoch": 1.61,
      "learning_rate": 3.849005424954793e-05,
      "loss": 0.2945,
      "step": 1273
    },
    {
      "epoch": 1.61,
      "learning_rate": 3.848101265822785e-05,
      "loss": 0.2304,
      "step": 1274
    },
    {
      "epoch": 1.61,
      "learning_rate": 3.847197106690778e-05,
      "loss": 0.1603,
      "step": 1275
    },
    {
      "epoch": 1.61,
      "learning_rate": 3.846292947558771e-05,
      "loss": 0.0787,
      "step": 1276
    },
    {
      "epoch": 1.62,
      "learning_rate": 3.8453887884267634e-05,
      "loss": 0.1269,
      "step": 1277
    },
    {
      "epoch": 1.62,
      "learning_rate": 3.844484629294756e-05,
      "loss": 0.2989,
      "step": 1278
    },
    {
      "epoch": 1.62,
      "learning_rate": 3.843580470162749e-05,
      "loss": 0.1222,
      "step": 1279
    },
    {
      "epoch": 1.62,
      "learning_rate": 3.8426763110307415e-05,
      "loss": 0.3145,
      "step": 1280
    },
    {
      "epoch": 1.62,
      "learning_rate": 3.841772151898734e-05,
      "loss": 0.3435,
      "step": 1281
    },
    {
      "epoch": 1.62,
      "learning_rate": 3.840867992766727e-05,
      "loss": 0.1968,
      "step": 1282
    },
    {
      "epoch": 1.62,
      "learning_rate": 3.8399638336347196e-05,
      "loss": 0.2114,
      "step": 1283
    },
    {
      "epoch": 1.62,
      "learning_rate": 3.839059674502712e-05,
      "loss": 0.2007,
      "step": 1284
    },
    {
      "epoch": 1.63,
      "learning_rate": 3.838155515370705e-05,
      "loss": 0.2732,
      "step": 1285
    },
    {
      "epoch": 1.63,
      "learning_rate": 3.837251356238698e-05,
      "loss": 0.3859,
      "step": 1286
    },
    {
      "epoch": 1.63,
      "learning_rate": 3.836347197106691e-05,
      "loss": 0.2916,
      "step": 1287
    },
    {
      "epoch": 1.63,
      "learning_rate": 3.8354430379746834e-05,
      "loss": 0.1319,
      "step": 1288
    },
    {
      "epoch": 1.63,
      "learning_rate": 3.8345388788426766e-05,
      "loss": 0.1531,
      "step": 1289
    },
    {
      "epoch": 1.63,
      "learning_rate": 3.833634719710669e-05,
      "loss": 0.1046,
      "step": 1290
    },
    {
      "epoch": 1.63,
      "learning_rate": 3.8327305605786615e-05,
      "loss": 0.1218,
      "step": 1291
    },
    {
      "epoch": 1.64,
      "learning_rate": 3.831826401446655e-05,
      "loss": 0.2911,
      "step": 1292
    },
    {
      "epoch": 1.64,
      "learning_rate": 3.830922242314647e-05,
      "loss": 0.0701,
      "step": 1293
    },
    {
      "epoch": 1.64,
      "learning_rate": 3.8300180831826404e-05,
      "loss": 0.3379,
      "step": 1294
    },
    {
      "epoch": 1.64,
      "learning_rate": 3.829113924050633e-05,
      "loss": 0.2034,
      "step": 1295
    },
    {
      "epoch": 1.64,
      "learning_rate": 3.828209764918626e-05,
      "loss": 0.1279,
      "step": 1296
    },
    {
      "epoch": 1.64,
      "learning_rate": 3.8273056057866185e-05,
      "loss": 0.2517,
      "step": 1297
    },
    {
      "epoch": 1.64,
      "learning_rate": 3.8264014466546117e-05,
      "loss": 0.2502,
      "step": 1298
    },
    {
      "epoch": 1.64,
      "learning_rate": 3.825497287522604e-05,
      "loss": 0.0457,
      "step": 1299
    },
    {
      "epoch": 1.65,
      "learning_rate": 3.8245931283905966e-05,
      "loss": 0.1933,
      "step": 1300
    },
    {
      "epoch": 1.65,
      "learning_rate": 3.82368896925859e-05,
      "loss": 0.1256,
      "step": 1301
    },
    {
      "epoch": 1.65,
      "learning_rate": 3.822784810126582e-05,
      "loss": 0.2278,
      "step": 1302
    },
    {
      "epoch": 1.65,
      "learning_rate": 3.8218806509945754e-05,
      "loss": 0.1178,
      "step": 1303
    },
    {
      "epoch": 1.65,
      "learning_rate": 3.820976491862568e-05,
      "loss": 0.1018,
      "step": 1304
    },
    {
      "epoch": 1.65,
      "learning_rate": 3.820072332730561e-05,
      "loss": 0.0862,
      "step": 1305
    },
    {
      "epoch": 1.65,
      "learning_rate": 3.8191681735985536e-05,
      "loss": 0.3528,
      "step": 1306
    },
    {
      "epoch": 1.65,
      "learning_rate": 3.818264014466546e-05,
      "loss": 0.326,
      "step": 1307
    },
    {
      "epoch": 1.66,
      "learning_rate": 3.817359855334539e-05,
      "loss": 0.0868,
      "step": 1308
    },
    {
      "epoch": 1.66,
      "learning_rate": 3.816455696202532e-05,
      "loss": 0.2975,
      "step": 1309
    },
    {
      "epoch": 1.66,
      "learning_rate": 3.815551537070525e-05,
      "loss": 0.1288,
      "step": 1310
    },
    {
      "epoch": 1.66,
      "learning_rate": 3.8146473779385173e-05,
      "loss": 0.2603,
      "step": 1311
    },
    {
      "epoch": 1.66,
      "learning_rate": 3.8137432188065105e-05,
      "loss": 0.2743,
      "step": 1312
    },
    {
      "epoch": 1.66,
      "learning_rate": 3.812839059674503e-05,
      "loss": 0.0603,
      "step": 1313
    },
    {
      "epoch": 1.66,
      "learning_rate": 3.811934900542496e-05,
      "loss": 0.2379,
      "step": 1314
    },
    {
      "epoch": 1.66,
      "learning_rate": 3.8110307414104886e-05,
      "loss": 0.292,
      "step": 1315
    },
    {
      "epoch": 1.67,
      "learning_rate": 3.810126582278481e-05,
      "loss": 0.2464,
      "step": 1316
    },
    {
      "epoch": 1.67,
      "learning_rate": 3.809222423146474e-05,
      "loss": 0.2194,
      "step": 1317
    },
    {
      "epoch": 1.67,
      "learning_rate": 3.808318264014467e-05,
      "loss": 0.2065,
      "step": 1318
    },
    {
      "epoch": 1.67,
      "learning_rate": 3.80741410488246e-05,
      "loss": 0.1826,
      "step": 1319
    },
    {
      "epoch": 1.67,
      "learning_rate": 3.8065099457504524e-05,
      "loss": 0.3546,
      "step": 1320
    },
    {
      "epoch": 1.67,
      "learning_rate": 3.8056057866184456e-05,
      "loss": 0.1504,
      "step": 1321
    },
    {
      "epoch": 1.67,
      "learning_rate": 3.804701627486438e-05,
      "loss": 0.1353,
      "step": 1322
    },
    {
      "epoch": 1.67,
      "learning_rate": 3.8037974683544306e-05,
      "loss": 0.1636,
      "step": 1323
    },
    {
      "epoch": 1.68,
      "learning_rate": 3.802893309222423e-05,
      "loss": 0.1371,
      "step": 1324
    },
    {
      "epoch": 1.68,
      "learning_rate": 3.801989150090416e-05,
      "loss": 0.3301,
      "step": 1325
    },
    {
      "epoch": 1.68,
      "learning_rate": 3.801084990958409e-05,
      "loss": 0.1264,
      "step": 1326
    },
    {
      "epoch": 1.68,
      "learning_rate": 3.800180831826401e-05,
      "loss": 0.2629,
      "step": 1327
    },
    {
      "epoch": 1.68,
      "learning_rate": 3.799276672694394e-05,
      "loss": 0.0735,
      "step": 1328
    },
    {
      "epoch": 1.68,
      "learning_rate": 3.798372513562387e-05,
      "loss": 0.53,
      "step": 1329
    },
    {
      "epoch": 1.68,
      "learning_rate": 3.79746835443038e-05,
      "loss": 0.2273,
      "step": 1330
    },
    {
      "epoch": 1.68,
      "learning_rate": 3.7965641952983725e-05,
      "loss": 0.2454,
      "step": 1331
    },
    {
      "epoch": 1.69,
      "learning_rate": 3.795660036166365e-05,
      "loss": 0.089,
      "step": 1332
    },
    {
      "epoch": 1.69,
      "learning_rate": 3.794755877034358e-05,
      "loss": 0.0576,
      "step": 1333
    },
    {
      "epoch": 1.69,
      "learning_rate": 3.7938517179023506e-05,
      "loss": 0.2358,
      "step": 1334
    },
    {
      "epoch": 1.69,
      "learning_rate": 3.792947558770344e-05,
      "loss": 0.4532,
      "step": 1335
    },
    {
      "epoch": 1.69,
      "learning_rate": 3.792043399638336e-05,
      "loss": 0.4574,
      "step": 1336
    },
    {
      "epoch": 1.69,
      "learning_rate": 3.7911392405063294e-05,
      "loss": 0.0879,
      "step": 1337
    },
    {
      "epoch": 1.69,
      "learning_rate": 3.790235081374322e-05,
      "loss": 0.2514,
      "step": 1338
    },
    {
      "epoch": 1.69,
      "learning_rate": 3.7893309222423144e-05,
      "loss": 0.2138,
      "step": 1339
    },
    {
      "epoch": 1.7,
      "learning_rate": 3.7884267631103075e-05,
      "loss": 0.1872,
      "step": 1340
    },
    {
      "epoch": 1.7,
      "learning_rate": 3.7875226039783e-05,
      "loss": 0.3117,
      "step": 1341
    },
    {
      "epoch": 1.7,
      "learning_rate": 3.786618444846293e-05,
      "loss": 0.1964,
      "step": 1342
    },
    {
      "epoch": 1.7,
      "learning_rate": 3.785714285714286e-05,
      "loss": 0.229,
      "step": 1343
    },
    {
      "epoch": 1.7,
      "learning_rate": 3.784810126582279e-05,
      "loss": 0.2974,
      "step": 1344
    },
    {
      "epoch": 1.7,
      "learning_rate": 3.783905967450271e-05,
      "loss": 0.1216,
      "step": 1345
    },
    {
      "epoch": 1.7,
      "learning_rate": 3.7830018083182645e-05,
      "loss": 0.2302,
      "step": 1346
    },
    {
      "epoch": 1.7,
      "learning_rate": 3.782097649186257e-05,
      "loss": 0.2009,
      "step": 1347
    },
    {
      "epoch": 1.71,
      "learning_rate": 3.7811934900542495e-05,
      "loss": 0.1484,
      "step": 1348
    },
    {
      "epoch": 1.71,
      "learning_rate": 3.7802893309222426e-05,
      "loss": 0.2663,
      "step": 1349
    },
    {
      "epoch": 1.71,
      "learning_rate": 3.779385171790235e-05,
      "loss": 0.112,
      "step": 1350
    },
    {
      "epoch": 1.71,
      "learning_rate": 3.778481012658228e-05,
      "loss": 0.1115,
      "step": 1351
    },
    {
      "epoch": 1.71,
      "learning_rate": 3.777576853526221e-05,
      "loss": 0.1515,
      "step": 1352
    },
    {
      "epoch": 1.71,
      "learning_rate": 3.776672694394214e-05,
      "loss": 0.138,
      "step": 1353
    },
    {
      "epoch": 1.71,
      "learning_rate": 3.7757685352622064e-05,
      "loss": 0.0867,
      "step": 1354
    },
    {
      "epoch": 1.71,
      "learning_rate": 3.774864376130199e-05,
      "loss": 0.2577,
      "step": 1355
    },
    {
      "epoch": 1.72,
      "learning_rate": 3.773960216998192e-05,
      "loss": 0.244,
      "step": 1356
    },
    {
      "epoch": 1.72,
      "learning_rate": 3.7730560578661845e-05,
      "loss": 0.0948,
      "step": 1357
    },
    {
      "epoch": 1.72,
      "learning_rate": 3.772151898734178e-05,
      "loss": 0.6216,
      "step": 1358
    },
    {
      "epoch": 1.72,
      "learning_rate": 3.77124773960217e-05,
      "loss": 0.0903,
      "step": 1359
    },
    {
      "epoch": 1.72,
      "learning_rate": 3.770343580470163e-05,
      "loss": 0.1984,
      "step": 1360
    },
    {
      "epoch": 1.72,
      "learning_rate": 3.769439421338156e-05,
      "loss": 0.287,
      "step": 1361
    },
    {
      "epoch": 1.72,
      "learning_rate": 3.768535262206149e-05,
      "loss": 0.0786,
      "step": 1362
    },
    {
      "epoch": 1.72,
      "learning_rate": 3.7676311030741415e-05,
      "loss": 0.2947,
      "step": 1363
    },
    {
      "epoch": 1.73,
      "learning_rate": 3.766726943942134e-05,
      "loss": 0.0923,
      "step": 1364
    },
    {
      "epoch": 1.73,
      "learning_rate": 3.765822784810127e-05,
      "loss": 0.1179,
      "step": 1365
    },
    {
      "epoch": 1.73,
      "learning_rate": 3.7649186256781196e-05,
      "loss": 0.2272,
      "step": 1366
    },
    {
      "epoch": 1.73,
      "learning_rate": 3.764014466546113e-05,
      "loss": 0.1933,
      "step": 1367
    },
    {
      "epoch": 1.73,
      "learning_rate": 3.763110307414105e-05,
      "loss": 0.1283,
      "step": 1368
    },
    {
      "epoch": 1.73,
      "learning_rate": 3.762206148282098e-05,
      "loss": 0.1802,
      "step": 1369
    },
    {
      "epoch": 1.73,
      "learning_rate": 3.761301989150091e-05,
      "loss": 0.2261,
      "step": 1370
    },
    {
      "epoch": 1.73,
      "learning_rate": 3.7603978300180834e-05,
      "loss": 0.2282,
      "step": 1371
    },
    {
      "epoch": 1.74,
      "learning_rate": 3.759493670886076e-05,
      "loss": 0.4486,
      "step": 1372
    },
    {
      "epoch": 1.74,
      "learning_rate": 3.758589511754069e-05,
      "loss": 0.0898,
      "step": 1373
    },
    {
      "epoch": 1.74,
      "learning_rate": 3.7576853526220615e-05,
      "loss": 0.0883,
      "step": 1374
    },
    {
      "epoch": 1.74,
      "learning_rate": 3.756781193490054e-05,
      "loss": 0.1563,
      "step": 1375
    },
    {
      "epoch": 1.74,
      "learning_rate": 3.755877034358047e-05,
      "loss": 0.0535,
      "step": 1376
    },
    {
      "epoch": 1.74,
      "learning_rate": 3.7549728752260396e-05,
      "loss": 0.1417,
      "step": 1377
    },
    {
      "epoch": 1.74,
      "learning_rate": 3.754068716094033e-05,
      "loss": 0.1756,
      "step": 1378
    },
    {
      "epoch": 1.75,
      "learning_rate": 3.753164556962025e-05,
      "loss": 0.3172,
      "step": 1379
    },
    {
      "epoch": 1.75,
      "learning_rate": 3.752260397830018e-05,
      "loss": 0.1457,
      "step": 1380
    },
    {
      "epoch": 1.75,
      "learning_rate": 3.751356238698011e-05,
      "loss": 0.106,
      "step": 1381
    },
    {
      "epoch": 1.75,
      "learning_rate": 3.7504520795660034e-05,
      "loss": 0.1528,
      "step": 1382
    },
    {
      "epoch": 1.75,
      "learning_rate": 3.7495479204339966e-05,
      "loss": 0.127,
      "step": 1383
    },
    {
      "epoch": 1.75,
      "learning_rate": 3.748643761301989e-05,
      "loss": 0.0773,
      "step": 1384
    },
    {
      "epoch": 1.75,
      "learning_rate": 3.747739602169982e-05,
      "loss": 0.1319,
      "step": 1385
    },
    {
      "epoch": 1.75,
      "learning_rate": 3.746835443037975e-05,
      "loss": 0.0848,
      "step": 1386
    },
    {
      "epoch": 1.76,
      "learning_rate": 3.745931283905967e-05,
      "loss": 0.1674,
      "step": 1387
    },
    {
      "epoch": 1.76,
      "learning_rate": 3.7450271247739604e-05,
      "loss": 0.2137,
      "step": 1388
    },
    {
      "epoch": 1.76,
      "learning_rate": 3.744122965641953e-05,
      "loss": 0.0644,
      "step": 1389
    },
    {
      "epoch": 1.76,
      "learning_rate": 3.743218806509946e-05,
      "loss": 0.2478,
      "step": 1390
    },
    {
      "epoch": 1.76,
      "learning_rate": 3.7423146473779385e-05,
      "loss": 0.0603,
      "step": 1391
    },
    {
      "epoch": 1.76,
      "learning_rate": 3.741410488245932e-05,
      "loss": 0.1451,
      "step": 1392
    },
    {
      "epoch": 1.76,
      "learning_rate": 3.740506329113924e-05,
      "loss": 0.1274,
      "step": 1393
    },
    {
      "epoch": 1.76,
      "learning_rate": 3.739602169981917e-05,
      "loss": 0.342,
      "step": 1394
    },
    {
      "epoch": 1.77,
      "learning_rate": 3.73869801084991e-05,
      "loss": 0.1056,
      "step": 1395
    },
    {
      "epoch": 1.77,
      "learning_rate": 3.737793851717902e-05,
      "loss": 0.1206,
      "step": 1396
    },
    {
      "epoch": 1.77,
      "learning_rate": 3.7368896925858954e-05,
      "loss": 0.4063,
      "step": 1397
    },
    {
      "epoch": 1.77,
      "learning_rate": 3.735985533453888e-05,
      "loss": 0.2091,
      "step": 1398
    },
    {
      "epoch": 1.77,
      "learning_rate": 3.735081374321881e-05,
      "loss": 0.3075,
      "step": 1399
    },
    {
      "epoch": 1.77,
      "learning_rate": 3.7341772151898736e-05,
      "loss": 0.0837,
      "step": 1400
    },
    {
      "epoch": 1.77,
      "learning_rate": 3.733273056057867e-05,
      "loss": 0.3355,
      "step": 1401
    },
    {
      "epoch": 1.77,
      "learning_rate": 3.732368896925859e-05,
      "loss": 0.1478,
      "step": 1402
    },
    {
      "epoch": 1.78,
      "learning_rate": 3.731464737793852e-05,
      "loss": 0.2184,
      "step": 1403
    },
    {
      "epoch": 1.78,
      "learning_rate": 3.730560578661845e-05,
      "loss": 0.2998,
      "step": 1404
    },
    {
      "epoch": 1.78,
      "learning_rate": 3.7296564195298374e-05,
      "loss": 0.1899,
      "step": 1405
    },
    {
      "epoch": 1.78,
      "learning_rate": 3.7287522603978305e-05,
      "loss": 0.047,
      "step": 1406
    },
    {
      "epoch": 1.78,
      "learning_rate": 3.727848101265823e-05,
      "loss": 0.2077,
      "step": 1407
    },
    {
      "epoch": 1.78,
      "learning_rate": 3.726943942133816e-05,
      "loss": 0.3134,
      "step": 1408
    },
    {
      "epoch": 1.78,
      "learning_rate": 3.7260397830018087e-05,
      "loss": 0.1034,
      "step": 1409
    },
    {
      "epoch": 1.78,
      "learning_rate": 3.725135623869802e-05,
      "loss": 0.1252,
      "step": 1410
    },
    {
      "epoch": 1.79,
      "learning_rate": 3.724231464737794e-05,
      "loss": 0.0979,
      "step": 1411
    },
    {
      "epoch": 1.79,
      "learning_rate": 3.723327305605787e-05,
      "loss": 0.0689,
      "step": 1412
    },
    {
      "epoch": 1.79,
      "learning_rate": 3.72242314647378e-05,
      "loss": 0.2709,
      "step": 1413
    },
    {
      "epoch": 1.79,
      "learning_rate": 3.7215189873417724e-05,
      "loss": 0.1802,
      "step": 1414
    },
    {
      "epoch": 1.79,
      "learning_rate": 3.720614828209765e-05,
      "loss": 0.1398,
      "step": 1415
    },
    {
      "epoch": 1.79,
      "learning_rate": 3.719710669077758e-05,
      "loss": 0.1756,
      "step": 1416
    },
    {
      "epoch": 1.79,
      "learning_rate": 3.7188065099457506e-05,
      "loss": 0.1009,
      "step": 1417
    },
    {
      "epoch": 1.79,
      "learning_rate": 3.717902350813743e-05,
      "loss": 0.1536,
      "step": 1418
    },
    {
      "epoch": 1.8,
      "learning_rate": 3.716998191681736e-05,
      "loss": 0.0518,
      "step": 1419
    },
    {
      "epoch": 1.8,
      "learning_rate": 3.716094032549729e-05,
      "loss": 0.2561,
      "step": 1420
    },
    {
      "epoch": 1.8,
      "learning_rate": 3.715189873417721e-05,
      "loss": 0.15,
      "step": 1421
    },
    {
      "epoch": 1.8,
      "learning_rate": 3.7142857142857143e-05,
      "loss": 0.6535,
      "step": 1422
    },
    {
      "epoch": 1.8,
      "learning_rate": 3.713381555153707e-05,
      "loss": 0.1221,
      "step": 1423
    },
    {
      "epoch": 1.8,
      "learning_rate": 3.7124773960217e-05,
      "loss": 0.2498,
      "step": 1424
    },
    {
      "epoch": 1.8,
      "learning_rate": 3.7115732368896925e-05,
      "loss": 0.02,
      "step": 1425
    },
    {
      "epoch": 1.8,
      "learning_rate": 3.7106690777576856e-05,
      "loss": 0.2478,
      "step": 1426
    },
    {
      "epoch": 1.81,
      "learning_rate": 3.709764918625678e-05,
      "loss": 0.229,
      "step": 1427
    },
    {
      "epoch": 1.81,
      "learning_rate": 3.7088607594936706e-05,
      "loss": 0.0313,
      "step": 1428
    },
    {
      "epoch": 1.81,
      "learning_rate": 3.707956600361664e-05,
      "loss": 0.0941,
      "step": 1429
    },
    {
      "epoch": 1.81,
      "learning_rate": 3.707052441229656e-05,
      "loss": 0.2485,
      "step": 1430
    },
    {
      "epoch": 1.81,
      "learning_rate": 3.7061482820976494e-05,
      "loss": 0.281,
      "step": 1431
    },
    {
      "epoch": 1.81,
      "learning_rate": 3.705244122965642e-05,
      "loss": 0.1499,
      "step": 1432
    },
    {
      "epoch": 1.81,
      "learning_rate": 3.704339963833635e-05,
      "loss": 0.0841,
      "step": 1433
    },
    {
      "epoch": 1.81,
      "learning_rate": 3.7034358047016276e-05,
      "loss": 0.1702,
      "step": 1434
    },
    {
      "epoch": 1.82,
      "learning_rate": 3.70253164556962e-05,
      "loss": 0.0796,
      "step": 1435
    },
    {
      "epoch": 1.82,
      "learning_rate": 3.701627486437613e-05,
      "loss": 0.0523,
      "step": 1436
    },
    {
      "epoch": 1.82,
      "learning_rate": 3.700723327305606e-05,
      "loss": 0.2839,
      "step": 1437
    },
    {
      "epoch": 1.82,
      "learning_rate": 3.699819168173599e-05,
      "loss": 0.0569,
      "step": 1438
    },
    {
      "epoch": 1.82,
      "learning_rate": 3.698915009041591e-05,
      "loss": 0.29,
      "step": 1439
    },
    {
      "epoch": 1.82,
      "learning_rate": 3.6980108499095845e-05,
      "loss": 0.2213,
      "step": 1440
    },
    {
      "epoch": 1.82,
      "learning_rate": 3.697106690777577e-05,
      "loss": 0.1938,
      "step": 1441
    },
    {
      "epoch": 1.82,
      "learning_rate": 3.69620253164557e-05,
      "loss": 0.219,
      "step": 1442
    },
    {
      "epoch": 1.83,
      "learning_rate": 3.6952983725135626e-05,
      "loss": 0.3945,
      "step": 1443
    },
    {
      "epoch": 1.83,
      "learning_rate": 3.694394213381555e-05,
      "loss": 0.1957,
      "step": 1444
    },
    {
      "epoch": 1.83,
      "learning_rate": 3.693490054249548e-05,
      "loss": 0.0398,
      "step": 1445
    },
    {
      "epoch": 1.83,
      "learning_rate": 3.692585895117541e-05,
      "loss": 0.3453,
      "step": 1446
    },
    {
      "epoch": 1.83,
      "learning_rate": 3.691681735985534e-05,
      "loss": 0.1946,
      "step": 1447
    },
    {
      "epoch": 1.83,
      "learning_rate": 3.6907775768535264e-05,
      "loss": 0.1472,
      "step": 1448
    },
    {
      "epoch": 1.83,
      "learning_rate": 3.6898734177215196e-05,
      "loss": 0.0983,
      "step": 1449
    },
    {
      "epoch": 1.83,
      "learning_rate": 3.688969258589512e-05,
      "loss": 0.1501,
      "step": 1450
    },
    {
      "epoch": 1.84,
      "learning_rate": 3.6880650994575045e-05,
      "loss": 0.2244,
      "step": 1451
    },
    {
      "epoch": 1.84,
      "learning_rate": 3.687160940325498e-05,
      "loss": 0.1817,
      "step": 1452
    },
    {
      "epoch": 1.84,
      "learning_rate": 3.68625678119349e-05,
      "loss": 0.2534,
      "step": 1453
    },
    {
      "epoch": 1.84,
      "learning_rate": 3.6853526220614834e-05,
      "loss": 0.5487,
      "step": 1454
    },
    {
      "epoch": 1.84,
      "learning_rate": 3.684448462929476e-05,
      "loss": 0.1363,
      "step": 1455
    },
    {
      "epoch": 1.84,
      "learning_rate": 3.683544303797469e-05,
      "loss": 0.1891,
      "step": 1456
    },
    {
      "epoch": 1.84,
      "learning_rate": 3.6826401446654615e-05,
      "loss": 0.3642,
      "step": 1457
    },
    {
      "epoch": 1.85,
      "learning_rate": 3.6817359855334546e-05,
      "loss": 0.3108,
      "step": 1458
    },
    {
      "epoch": 1.85,
      "learning_rate": 3.680831826401447e-05,
      "loss": 0.0317,
      "step": 1459
    },
    {
      "epoch": 1.85,
      "learning_rate": 3.6799276672694396e-05,
      "loss": 0.2466,
      "step": 1460
    },
    {
      "epoch": 1.85,
      "learning_rate": 3.679023508137433e-05,
      "loss": 0.1529,
      "step": 1461
    },
    {
      "epoch": 1.85,
      "learning_rate": 3.678119349005425e-05,
      "loss": 0.1364,
      "step": 1462
    },
    {
      "epoch": 1.85,
      "learning_rate": 3.677215189873418e-05,
      "loss": 0.3486,
      "step": 1463
    },
    {
      "epoch": 1.85,
      "learning_rate": 3.676311030741411e-05,
      "loss": 0.3546,
      "step": 1464
    },
    {
      "epoch": 1.85,
      "learning_rate": 3.6754068716094034e-05,
      "loss": 0.2115,
      "step": 1465
    },
    {
      "epoch": 1.86,
      "learning_rate": 3.674502712477396e-05,
      "loss": 0.1788,
      "step": 1466
    },
    {
      "epoch": 1.86,
      "learning_rate": 3.673598553345389e-05,
      "loss": 0.1515,
      "step": 1467
    },
    {
      "epoch": 1.86,
      "learning_rate": 3.6726943942133815e-05,
      "loss": 0.0988,
      "step": 1468
    },
    {
      "epoch": 1.86,
      "learning_rate": 3.671790235081374e-05,
      "loss": 0.2497,
      "step": 1469
    },
    {
      "epoch": 1.86,
      "learning_rate": 3.670886075949367e-05,
      "loss": 0.3142,
      "step": 1470
    },
    {
      "epoch": 1.86,
      "learning_rate": 3.66998191681736e-05,
      "loss": 0.2043,
      "step": 1471
    },
    {
      "epoch": 1.86,
      "learning_rate": 3.669077757685353e-05,
      "loss": 0.2365,
      "step": 1472
    },
    {
      "epoch": 1.86,
      "learning_rate": 3.668173598553345e-05,
      "loss": 0.1046,
      "step": 1473
    },
    {
      "epoch": 1.87,
      "learning_rate": 3.6672694394213385e-05,
      "loss": 0.1171,
      "step": 1474
    },
    {
      "epoch": 1.87,
      "learning_rate": 3.666365280289331e-05,
      "loss": 0.12,
      "step": 1475
    },
    {
      "epoch": 1.87,
      "learning_rate": 3.6654611211573234e-05,
      "loss": 0.1333,
      "step": 1476
    },
    {
      "epoch": 1.87,
      "learning_rate": 3.6645569620253166e-05,
      "loss": 0.3652,
      "step": 1477
    },
    {
      "epoch": 1.87,
      "learning_rate": 3.663652802893309e-05,
      "loss": 0.2198,
      "step": 1478
    },
    {
      "epoch": 1.87,
      "learning_rate": 3.662748643761302e-05,
      "loss": 0.3439,
      "step": 1479
    },
    {
      "epoch": 1.87,
      "learning_rate": 3.661844484629295e-05,
      "loss": 0.143,
      "step": 1480
    },
    {
      "epoch": 1.87,
      "learning_rate": 3.660940325497288e-05,
      "loss": 0.1927,
      "step": 1481
    },
    {
      "epoch": 1.88,
      "learning_rate": 3.6600361663652804e-05,
      "loss": 0.2002,
      "step": 1482
    },
    {
      "epoch": 1.88,
      "learning_rate": 3.659132007233273e-05,
      "loss": 0.325,
      "step": 1483
    },
    {
      "epoch": 1.88,
      "learning_rate": 3.658227848101266e-05,
      "loss": 0.1084,
      "step": 1484
    },
    {
      "epoch": 1.88,
      "learning_rate": 3.6573236889692585e-05,
      "loss": 0.0642,
      "step": 1485
    },
    {
      "epoch": 1.88,
      "learning_rate": 3.656419529837252e-05,
      "loss": 0.1224,
      "step": 1486
    },
    {
      "epoch": 1.88,
      "learning_rate": 3.655515370705244e-05,
      "loss": 0.1135,
      "step": 1487
    },
    {
      "epoch": 1.88,
      "learning_rate": 3.654611211573237e-05,
      "loss": 0.0614,
      "step": 1488
    },
    {
      "epoch": 1.88,
      "learning_rate": 3.65370705244123e-05,
      "loss": 0.3241,
      "step": 1489
    },
    {
      "epoch": 1.89,
      "learning_rate": 3.652802893309223e-05,
      "loss": 0.4158,
      "step": 1490
    },
    {
      "epoch": 1.89,
      "learning_rate": 3.6518987341772155e-05,
      "loss": 0.5297,
      "step": 1491
    },
    {
      "epoch": 1.89,
      "learning_rate": 3.650994575045208e-05,
      "loss": 0.1484,
      "step": 1492
    },
    {
      "epoch": 1.89,
      "learning_rate": 3.650090415913201e-05,
      "loss": 0.562,
      "step": 1493
    },
    {
      "epoch": 1.89,
      "learning_rate": 3.6491862567811936e-05,
      "loss": 0.3228,
      "step": 1494
    },
    {
      "epoch": 1.89,
      "learning_rate": 3.648282097649187e-05,
      "loss": 0.3775,
      "step": 1495
    },
    {
      "epoch": 1.89,
      "learning_rate": 3.647377938517179e-05,
      "loss": 0.1366,
      "step": 1496
    },
    {
      "epoch": 1.89,
      "learning_rate": 3.6464737793851724e-05,
      "loss": 0.3189,
      "step": 1497
    },
    {
      "epoch": 1.9,
      "learning_rate": 3.645569620253165e-05,
      "loss": 0.0709,
      "step": 1498
    },
    {
      "epoch": 1.9,
      "learning_rate": 3.6446654611211574e-05,
      "loss": 0.4296,
      "step": 1499
    },
    {
      "epoch": 1.9,
      "learning_rate": 3.6437613019891505e-05,
      "loss": 0.365,
      "step": 1500
    },
    {
      "epoch": 1.9,
      "eval_accuracy": 0.9091591591591591,
      "eval_loss": 0.22769701480865479,
      "eval_runtime": 45.14,
      "eval_samples_per_second": 29.508,
      "eval_steps_per_second": 3.7,
      "step": 1500
    },
    {
      "epoch": 1.9,
      "learning_rate": 3.642857142857143e-05,
      "loss": 0.1703,
      "step": 1501
    },
    {
      "epoch": 1.9,
      "learning_rate": 3.641952983725136e-05,
      "loss": 0.2375,
      "step": 1502
    },
    {
      "epoch": 1.9,
      "learning_rate": 3.641048824593129e-05,
      "loss": 0.2937,
      "step": 1503
    },
    {
      "epoch": 1.9,
      "learning_rate": 3.640144665461122e-05,
      "loss": 0.2186,
      "step": 1504
    },
    {
      "epoch": 1.9,
      "learning_rate": 3.639240506329114e-05,
      "loss": 0.0876,
      "step": 1505
    },
    {
      "epoch": 1.91,
      "learning_rate": 3.6383363471971075e-05,
      "loss": 0.1912,
      "step": 1506
    },
    {
      "epoch": 1.91,
      "learning_rate": 3.6374321880651e-05,
      "loss": 0.3821,
      "step": 1507
    },
    {
      "epoch": 1.91,
      "learning_rate": 3.6365280289330924e-05,
      "loss": 0.1691,
      "step": 1508
    },
    {
      "epoch": 1.91,
      "learning_rate": 3.635623869801085e-05,
      "loss": 0.1933,
      "step": 1509
    },
    {
      "epoch": 1.91,
      "learning_rate": 3.634719710669078e-05,
      "loss": 0.4963,
      "step": 1510
    },
    {
      "epoch": 1.91,
      "learning_rate": 3.6338155515370706e-05,
      "loss": 0.212,
      "step": 1511
    },
    {
      "epoch": 1.91,
      "learning_rate": 3.632911392405063e-05,
      "loss": 0.1115,
      "step": 1512
    },
    {
      "epoch": 1.91,
      "learning_rate": 3.632007233273056e-05,
      "loss": 0.0585,
      "step": 1513
    },
    {
      "epoch": 1.92,
      "learning_rate": 3.631103074141049e-05,
      "loss": 0.1556,
      "step": 1514
    },
    {
      "epoch": 1.92,
      "learning_rate": 3.630198915009041e-05,
      "loss": 0.2808,
      "step": 1515
    },
    {
      "epoch": 1.92,
      "learning_rate": 3.6292947558770344e-05,
      "loss": 0.1247,
      "step": 1516
    },
    {
      "epoch": 1.92,
      "learning_rate": 3.628390596745027e-05,
      "loss": 0.1466,
      "step": 1517
    },
    {
      "epoch": 1.92,
      "learning_rate": 3.62748643761302e-05,
      "loss": 0.2439,
      "step": 1518
    },
    {
      "epoch": 1.92,
      "learning_rate": 3.6265822784810125e-05,
      "loss": 0.2864,
      "step": 1519
    },
    {
      "epoch": 1.92,
      "learning_rate": 3.6256781193490057e-05,
      "loss": 0.4481,
      "step": 1520
    },
    {
      "epoch": 1.92,
      "learning_rate": 3.624773960216998e-05,
      "loss": 0.2471,
      "step": 1521
    },
    {
      "epoch": 1.93,
      "learning_rate": 3.6238698010849906e-05,
      "loss": 0.1669,
      "step": 1522
    },
    {
      "epoch": 1.93,
      "learning_rate": 3.622965641952984e-05,
      "loss": 0.1775,
      "step": 1523
    },
    {
      "epoch": 1.93,
      "learning_rate": 3.622061482820976e-05,
      "loss": 0.2219,
      "step": 1524
    },
    {
      "epoch": 1.93,
      "learning_rate": 3.6211573236889694e-05,
      "loss": 0.1348,
      "step": 1525
    },
    {
      "epoch": 1.93,
      "learning_rate": 3.620253164556962e-05,
      "loss": 0.1507,
      "step": 1526
    },
    {
      "epoch": 1.93,
      "learning_rate": 3.619349005424955e-05,
      "loss": 0.2031,
      "step": 1527
    },
    {
      "epoch": 1.93,
      "learning_rate": 3.6184448462929476e-05,
      "loss": 0.0885,
      "step": 1528
    },
    {
      "epoch": 1.93,
      "learning_rate": 3.617540687160941e-05,
      "loss": 0.3382,
      "step": 1529
    },
    {
      "epoch": 1.94,
      "learning_rate": 3.616636528028933e-05,
      "loss": 0.1977,
      "step": 1530
    },
    {
      "epoch": 1.94,
      "learning_rate": 3.615732368896926e-05,
      "loss": 0.2903,
      "step": 1531
    },
    {
      "epoch": 1.94,
      "learning_rate": 3.614828209764919e-05,
      "loss": 0.1289,
      "step": 1532
    },
    {
      "epoch": 1.94,
      "learning_rate": 3.6139240506329114e-05,
      "loss": 0.1809,
      "step": 1533
    },
    {
      "epoch": 1.94,
      "learning_rate": 3.6130198915009045e-05,
      "loss": 0.0965,
      "step": 1534
    },
    {
      "epoch": 1.94,
      "learning_rate": 3.612115732368897e-05,
      "loss": 0.1445,
      "step": 1535
    },
    {
      "epoch": 1.94,
      "learning_rate": 3.61121157323689e-05,
      "loss": 0.3118,
      "step": 1536
    },
    {
      "epoch": 1.94,
      "learning_rate": 3.6103074141048826e-05,
      "loss": 0.0676,
      "step": 1537
    },
    {
      "epoch": 1.95,
      "learning_rate": 3.609403254972875e-05,
      "loss": 0.288,
      "step": 1538
    },
    {
      "epoch": 1.95,
      "learning_rate": 3.608499095840868e-05,
      "loss": 0.1393,
      "step": 1539
    },
    {
      "epoch": 1.95,
      "learning_rate": 3.607594936708861e-05,
      "loss": 0.0616,
      "step": 1540
    },
    {
      "epoch": 1.95,
      "learning_rate": 3.606690777576854e-05,
      "loss": 0.3541,
      "step": 1541
    },
    {
      "epoch": 1.95,
      "learning_rate": 3.6057866184448464e-05,
      "loss": 0.3352,
      "step": 1542
    },
    {
      "epoch": 1.95,
      "learning_rate": 3.6048824593128396e-05,
      "loss": 0.2457,
      "step": 1543
    },
    {
      "epoch": 1.95,
      "learning_rate": 3.603978300180832e-05,
      "loss": 0.1275,
      "step": 1544
    },
    {
      "epoch": 1.96,
      "learning_rate": 3.603074141048825e-05,
      "loss": 0.2752,
      "step": 1545
    },
    {
      "epoch": 1.96,
      "learning_rate": 3.602169981916818e-05,
      "loss": 0.2895,
      "step": 1546
    },
    {
      "epoch": 1.96,
      "learning_rate": 3.60126582278481e-05,
      "loss": 0.6552,
      "step": 1547
    },
    {
      "epoch": 1.96,
      "learning_rate": 3.6003616636528034e-05,
      "loss": 0.3267,
      "step": 1548
    },
    {
      "epoch": 1.96,
      "learning_rate": 3.599457504520796e-05,
      "loss": 0.2243,
      "step": 1549
    },
    {
      "epoch": 1.96,
      "learning_rate": 3.598553345388789e-05,
      "loss": 0.2289,
      "step": 1550
    },
    {
      "epoch": 1.96,
      "learning_rate": 3.5976491862567815e-05,
      "loss": 0.1563,
      "step": 1551
    },
    {
      "epoch": 1.96,
      "learning_rate": 3.596745027124775e-05,
      "loss": 0.1291,
      "step": 1552
    },
    {
      "epoch": 1.97,
      "learning_rate": 3.595840867992767e-05,
      "loss": 0.1651,
      "step": 1553
    },
    {
      "epoch": 1.97,
      "learning_rate": 3.5949367088607596e-05,
      "loss": 0.2285,
      "step": 1554
    },
    {
      "epoch": 1.97,
      "learning_rate": 3.594032549728753e-05,
      "loss": 0.3111,
      "step": 1555
    },
    {
      "epoch": 1.97,
      "learning_rate": 3.593128390596745e-05,
      "loss": 0.1825,
      "step": 1556
    },
    {
      "epoch": 1.97,
      "learning_rate": 3.592224231464738e-05,
      "loss": 0.101,
      "step": 1557
    },
    {
      "epoch": 1.97,
      "learning_rate": 3.591320072332731e-05,
      "loss": 0.1259,
      "step": 1558
    },
    {
      "epoch": 1.97,
      "learning_rate": 3.5904159132007234e-05,
      "loss": 0.1314,
      "step": 1559
    },
    {
      "epoch": 1.97,
      "learning_rate": 3.589511754068716e-05,
      "loss": 0.0779,
      "step": 1560
    },
    {
      "epoch": 1.98,
      "learning_rate": 3.588607594936709e-05,
      "loss": 0.2323,
      "step": 1561
    },
    {
      "epoch": 1.98,
      "learning_rate": 3.5877034358047015e-05,
      "loss": 0.1958,
      "step": 1562
    },
    {
      "epoch": 1.98,
      "learning_rate": 3.586799276672694e-05,
      "loss": 0.2267,
      "step": 1563
    },
    {
      "epoch": 1.98,
      "learning_rate": 3.585895117540687e-05,
      "loss": 0.2488,
      "step": 1564
    },
    {
      "epoch": 1.98,
      "learning_rate": 3.58499095840868e-05,
      "loss": 0.2456,
      "step": 1565
    },
    {
      "epoch": 1.98,
      "learning_rate": 3.584086799276673e-05,
      "loss": 0.2111,
      "step": 1566
    },
    {
      "epoch": 1.98,
      "learning_rate": 3.583182640144665e-05,
      "loss": 0.2204,
      "step": 1567
    },
    {
      "epoch": 1.98,
      "learning_rate": 3.5822784810126585e-05,
      "loss": 0.2159,
      "step": 1568
    },
    {
      "epoch": 1.99,
      "learning_rate": 3.581374321880651e-05,
      "loss": 0.2913,
      "step": 1569
    },
    {
      "epoch": 1.99,
      "learning_rate": 3.5804701627486435e-05,
      "loss": 0.1309,
      "step": 1570
    },
    {
      "epoch": 1.99,
      "learning_rate": 3.5795660036166366e-05,
      "loss": 0.1102,
      "step": 1571
    },
    {
      "epoch": 1.99,
      "learning_rate": 3.578661844484629e-05,
      "loss": 0.1917,
      "step": 1572
    },
    {
      "epoch": 1.99,
      "learning_rate": 3.577757685352622e-05,
      "loss": 0.4332,
      "step": 1573
    },
    {
      "epoch": 1.99,
      "learning_rate": 3.576853526220615e-05,
      "loss": 0.207,
      "step": 1574
    },
    {
      "epoch": 1.99,
      "learning_rate": 3.575949367088608e-05,
      "loss": 0.2693,
      "step": 1575
    },
    {
      "epoch": 1.99,
      "learning_rate": 3.5750452079566004e-05,
      "loss": 0.2737,
      "step": 1576
    },
    {
      "epoch": 2.0,
      "learning_rate": 3.5741410488245936e-05,
      "loss": 0.3569,
      "step": 1577
    },
    {
      "epoch": 2.0,
      "learning_rate": 3.573236889692586e-05,
      "loss": 0.2986,
      "step": 1578
    },
    {
      "epoch": 2.0,
      "learning_rate": 3.5723327305605785e-05,
      "loss": 0.2576,
      "step": 1579
    },
    {
      "epoch": 2.0,
      "learning_rate": 3.571428571428572e-05,
      "loss": 0.2772,
      "step": 1580
    },
    {
      "epoch": 2.0,
      "learning_rate": 3.570524412296564e-05,
      "loss": 0.085,
      "step": 1581
    },
    {
      "epoch": 2.0,
      "learning_rate": 3.5696202531645573e-05,
      "loss": 0.3006,
      "step": 1582
    },
    {
      "epoch": 2.0,
      "learning_rate": 3.56871609403255e-05,
      "loss": 0.0506,
      "step": 1583
    },
    {
      "epoch": 2.01,
      "learning_rate": 3.567811934900543e-05,
      "loss": 0.0353,
      "step": 1584
    },
    {
      "epoch": 2.01,
      "learning_rate": 3.5669077757685355e-05,
      "loss": 0.0669,
      "step": 1585
    },
    {
      "epoch": 2.01,
      "learning_rate": 3.566003616636528e-05,
      "loss": 0.2118,
      "step": 1586
    },
    {
      "epoch": 2.01,
      "learning_rate": 3.565099457504521e-05,
      "loss": 0.1156,
      "step": 1587
    },
    {
      "epoch": 2.01,
      "learning_rate": 3.5641952983725136e-05,
      "loss": 0.1123,
      "step": 1588
    },
    {
      "epoch": 2.01,
      "learning_rate": 3.563291139240507e-05,
      "loss": 0.2213,
      "step": 1589
    },
    {
      "epoch": 2.01,
      "learning_rate": 3.562386980108499e-05,
      "loss": 0.1276,
      "step": 1590
    },
    {
      "epoch": 2.01,
      "learning_rate": 3.5614828209764924e-05,
      "loss": 0.096,
      "step": 1591
    },
    {
      "epoch": 2.02,
      "learning_rate": 3.560578661844485e-05,
      "loss": 0.1573,
      "step": 1592
    },
    {
      "epoch": 2.02,
      "learning_rate": 3.559674502712478e-05,
      "loss": 0.1087,
      "step": 1593
    },
    {
      "epoch": 2.02,
      "learning_rate": 3.5587703435804706e-05,
      "loss": 0.0952,
      "step": 1594
    },
    {
      "epoch": 2.02,
      "learning_rate": 3.557866184448463e-05,
      "loss": 0.1067,
      "step": 1595
    },
    {
      "epoch": 2.02,
      "learning_rate": 3.556962025316456e-05,
      "loss": 0.0974,
      "step": 1596
    },
    {
      "epoch": 2.02,
      "learning_rate": 3.556057866184449e-05,
      "loss": 0.0841,
      "step": 1597
    },
    {
      "epoch": 2.02,
      "learning_rate": 3.555153707052442e-05,
      "loss": 0.1074,
      "step": 1598
    },
    {
      "epoch": 2.02,
      "learning_rate": 3.554249547920434e-05,
      "loss": 0.0614,
      "step": 1599
    },
    {
      "epoch": 2.03,
      "learning_rate": 3.553345388788427e-05,
      "loss": 0.0408,
      "step": 1600
    },
    {
      "epoch": 2.03,
      "learning_rate": 3.55244122965642e-05,
      "loss": 0.1173,
      "step": 1601
    },
    {
      "epoch": 2.03,
      "learning_rate": 3.5515370705244125e-05,
      "loss": 0.0943,
      "step": 1602
    },
    {
      "epoch": 2.03,
      "learning_rate": 3.550632911392405e-05,
      "loss": 0.1234,
      "step": 1603
    },
    {
      "epoch": 2.03,
      "learning_rate": 3.549728752260398e-05,
      "loss": 0.0469,
      "step": 1604
    },
    {
      "epoch": 2.03,
      "learning_rate": 3.5488245931283906e-05,
      "loss": 0.1018,
      "step": 1605
    },
    {
      "epoch": 2.03,
      "learning_rate": 3.547920433996383e-05,
      "loss": 0.1139,
      "step": 1606
    },
    {
      "epoch": 2.03,
      "learning_rate": 3.547016274864376e-05,
      "loss": 0.1649,
      "step": 1607
    },
    {
      "epoch": 2.04,
      "learning_rate": 3.546112115732369e-05,
      "loss": 0.0783,
      "step": 1608
    },
    {
      "epoch": 2.04,
      "learning_rate": 3.545207956600362e-05,
      "loss": 0.1741,
      "step": 1609
    },
    {
      "epoch": 2.04,
      "learning_rate": 3.5443037974683544e-05,
      "loss": 0.0731,
      "step": 1610
    },
    {
      "epoch": 2.04,
      "learning_rate": 3.543399638336347e-05,
      "loss": 0.1451,
      "step": 1611
    },
    {
      "epoch": 2.04,
      "learning_rate": 3.54249547920434e-05,
      "loss": 0.2753,
      "step": 1612
    },
    {
      "epoch": 2.04,
      "learning_rate": 3.5415913200723325e-05,
      "loss": 0.1274,
      "step": 1613
    },
    {
      "epoch": 2.04,
      "learning_rate": 3.540687160940326e-05,
      "loss": 0.1052,
      "step": 1614
    },
    {
      "epoch": 2.04,
      "learning_rate": 3.539783001808318e-05,
      "loss": 0.2063,
      "step": 1615
    },
    {
      "epoch": 2.05,
      "learning_rate": 3.538878842676311e-05,
      "loss": 0.387,
      "step": 1616
    },
    {
      "epoch": 2.05,
      "learning_rate": 3.537974683544304e-05,
      "loss": 0.0245,
      "step": 1617
    },
    {
      "epoch": 2.05,
      "learning_rate": 3.537070524412296e-05,
      "loss": 0.2246,
      "step": 1618
    },
    {
      "epoch": 2.05,
      "learning_rate": 3.5361663652802895e-05,
      "loss": 0.0569,
      "step": 1619
    },
    {
      "epoch": 2.05,
      "learning_rate": 3.535262206148282e-05,
      "loss": 0.0419,
      "step": 1620
    },
    {
      "epoch": 2.05,
      "learning_rate": 3.534358047016275e-05,
      "loss": 0.0095,
      "step": 1621
    },
    {
      "epoch": 2.05,
      "learning_rate": 3.5334538878842676e-05,
      "loss": 0.1783,
      "step": 1622
    },
    {
      "epoch": 2.05,
      "learning_rate": 3.532549728752261e-05,
      "loss": 0.0311,
      "step": 1623
    },
    {
      "epoch": 2.06,
      "learning_rate": 3.531645569620253e-05,
      "loss": 0.1062,
      "step": 1624
    },
    {
      "epoch": 2.06,
      "learning_rate": 3.5307414104882464e-05,
      "loss": 0.1161,
      "step": 1625
    },
    {
      "epoch": 2.06,
      "learning_rate": 3.529837251356239e-05,
      "loss": 0.1821,
      "step": 1626
    },
    {
      "epoch": 2.06,
      "learning_rate": 3.5289330922242314e-05,
      "loss": 0.0855,
      "step": 1627
    },
    {
      "epoch": 2.06,
      "learning_rate": 3.5280289330922245e-05,
      "loss": 0.0377,
      "step": 1628
    },
    {
      "epoch": 2.06,
      "learning_rate": 3.527124773960217e-05,
      "loss": 0.0681,
      "step": 1629
    },
    {
      "epoch": 2.06,
      "learning_rate": 3.52622061482821e-05,
      "loss": 0.1032,
      "step": 1630
    },
    {
      "epoch": 2.06,
      "learning_rate": 3.5253164556962027e-05,
      "loss": 0.099,
      "step": 1631
    },
    {
      "epoch": 2.07,
      "learning_rate": 3.524412296564196e-05,
      "loss": 0.0181,
      "step": 1632
    },
    {
      "epoch": 2.07,
      "learning_rate": 3.523508137432188e-05,
      "loss": 0.0141,
      "step": 1633
    },
    {
      "epoch": 2.07,
      "learning_rate": 3.522603978300181e-05,
      "loss": 0.1853,
      "step": 1634
    },
    {
      "epoch": 2.07,
      "learning_rate": 3.521699819168174e-05,
      "loss": 0.3225,
      "step": 1635
    },
    {
      "epoch": 2.07,
      "learning_rate": 3.5207956600361664e-05,
      "loss": 0.1515,
      "step": 1636
    },
    {
      "epoch": 2.07,
      "learning_rate": 3.5198915009041596e-05,
      "loss": 0.0939,
      "step": 1637
    },
    {
      "epoch": 2.07,
      "learning_rate": 3.518987341772152e-05,
      "loss": 0.3555,
      "step": 1638
    },
    {
      "epoch": 2.07,
      "learning_rate": 3.518083182640145e-05,
      "loss": 0.2026,
      "step": 1639
    },
    {
      "epoch": 2.08,
      "learning_rate": 3.517179023508138e-05,
      "loss": 0.0237,
      "step": 1640
    },
    {
      "epoch": 2.08,
      "learning_rate": 3.516274864376131e-05,
      "loss": 0.2579,
      "step": 1641
    },
    {
      "epoch": 2.08,
      "learning_rate": 3.5153707052441234e-05,
      "loss": 0.0154,
      "step": 1642
    },
    {
      "epoch": 2.08,
      "learning_rate": 3.514466546112116e-05,
      "loss": 0.3121,
      "step": 1643
    },
    {
      "epoch": 2.08,
      "learning_rate": 3.513562386980109e-05,
      "loss": 0.0214,
      "step": 1644
    },
    {
      "epoch": 2.08,
      "learning_rate": 3.5126582278481015e-05,
      "loss": 0.1827,
      "step": 1645
    },
    {
      "epoch": 2.08,
      "learning_rate": 3.511754068716095e-05,
      "loss": 0.042,
      "step": 1646
    },
    {
      "epoch": 2.08,
      "learning_rate": 3.510849909584087e-05,
      "loss": 0.0147,
      "step": 1647
    },
    {
      "epoch": 2.09,
      "learning_rate": 3.5099457504520796e-05,
      "loss": 0.0501,
      "step": 1648
    },
    {
      "epoch": 2.09,
      "learning_rate": 3.509041591320073e-05,
      "loss": 0.1335,
      "step": 1649
    },
    {
      "epoch": 2.09,
      "learning_rate": 3.508137432188065e-05,
      "loss": 0.0219,
      "step": 1650
    },
    {
      "epoch": 2.09,
      "learning_rate": 3.507233273056058e-05,
      "loss": 0.075,
      "step": 1651
    },
    {
      "epoch": 2.09,
      "learning_rate": 3.506329113924051e-05,
      "loss": 0.1293,
      "step": 1652
    },
    {
      "epoch": 2.09,
      "learning_rate": 3.5054249547920434e-05,
      "loss": 0.0269,
      "step": 1653
    },
    {
      "epoch": 2.09,
      "learning_rate": 3.504520795660036e-05,
      "loss": 0.1505,
      "step": 1654
    },
    {
      "epoch": 2.09,
      "learning_rate": 3.503616636528029e-05,
      "loss": 0.0552,
      "step": 1655
    },
    {
      "epoch": 2.1,
      "learning_rate": 3.5027124773960216e-05,
      "loss": 0.0508,
      "step": 1656
    },
    {
      "epoch": 2.1,
      "learning_rate": 3.501808318264015e-05,
      "loss": 0.1009,
      "step": 1657
    },
    {
      "epoch": 2.1,
      "learning_rate": 3.500904159132007e-05,
      "loss": 0.1671,
      "step": 1658
    },
    {
      "epoch": 2.1,
      "learning_rate": 3.5e-05,
      "loss": 0.0632,
      "step": 1659
    },
    {
      "epoch": 2.1,
      "learning_rate": 3.499095840867993e-05,
      "loss": 0.0771,
      "step": 1660
    },
    {
      "epoch": 2.1,
      "learning_rate": 3.4981916817359853e-05,
      "loss": 0.2198,
      "step": 1661
    },
    {
      "epoch": 2.1,
      "learning_rate": 3.4972875226039785e-05,
      "loss": 0.0686,
      "step": 1662
    },
    {
      "epoch": 2.1,
      "learning_rate": 3.496383363471971e-05,
      "loss": 0.1076,
      "step": 1663
    },
    {
      "epoch": 2.11,
      "learning_rate": 3.495479204339964e-05,
      "loss": 0.1646,
      "step": 1664
    },
    {
      "epoch": 2.11,
      "learning_rate": 3.4945750452079566e-05,
      "loss": 0.0679,
      "step": 1665
    },
    {
      "epoch": 2.11,
      "learning_rate": 3.493670886075949e-05,
      "loss": 0.0472,
      "step": 1666
    },
    {
      "epoch": 2.11,
      "learning_rate": 3.492766726943942e-05,
      "loss": 0.065,
      "step": 1667
    },
    {
      "epoch": 2.11,
      "learning_rate": 3.491862567811935e-05,
      "loss": 0.0696,
      "step": 1668
    },
    {
      "epoch": 2.11,
      "learning_rate": 3.490958408679928e-05,
      "loss": 0.0773,
      "step": 1669
    },
    {
      "epoch": 2.11,
      "learning_rate": 3.4900542495479204e-05,
      "loss": 0.1291,
      "step": 1670
    },
    {
      "epoch": 2.12,
      "learning_rate": 3.4891500904159136e-05,
      "loss": 0.0787,
      "step": 1671
    },
    {
      "epoch": 2.12,
      "learning_rate": 3.488245931283906e-05,
      "loss": 0.0469,
      "step": 1672
    },
    {
      "epoch": 2.12,
      "learning_rate": 3.487341772151899e-05,
      "loss": 0.0512,
      "step": 1673
    },
    {
      "epoch": 2.12,
      "learning_rate": 3.486437613019892e-05,
      "loss": 0.1399,
      "step": 1674
    },
    {
      "epoch": 2.12,
      "learning_rate": 3.485533453887884e-05,
      "loss": 0.0988,
      "step": 1675
    },
    {
      "epoch": 2.12,
      "learning_rate": 3.4846292947558774e-05,
      "loss": 0.0724,
      "step": 1676
    },
    {
      "epoch": 2.12,
      "learning_rate": 3.48372513562387e-05,
      "loss": 0.0218,
      "step": 1677
    },
    {
      "epoch": 2.12,
      "learning_rate": 3.482820976491863e-05,
      "loss": 0.0967,
      "step": 1678
    },
    {
      "epoch": 2.13,
      "learning_rate": 3.4819168173598555e-05,
      "loss": 0.3307,
      "step": 1679
    },
    {
      "epoch": 2.13,
      "learning_rate": 3.4810126582278487e-05,
      "loss": 0.2752,
      "step": 1680
    },
    {
      "epoch": 2.13,
      "learning_rate": 3.480108499095841e-05,
      "loss": 0.2112,
      "step": 1681
    },
    {
      "epoch": 2.13,
      "learning_rate": 3.4792043399638336e-05,
      "loss": 0.0898,
      "step": 1682
    },
    {
      "epoch": 2.13,
      "learning_rate": 3.478300180831827e-05,
      "loss": 0.1622,
      "step": 1683
    },
    {
      "epoch": 2.13,
      "learning_rate": 3.477396021699819e-05,
      "loss": 0.0641,
      "step": 1684
    },
    {
      "epoch": 2.13,
      "learning_rate": 3.4764918625678124e-05,
      "loss": 0.4007,
      "step": 1685
    },
    {
      "epoch": 2.13,
      "learning_rate": 3.475587703435805e-05,
      "loss": 0.0304,
      "step": 1686
    },
    {
      "epoch": 2.14,
      "learning_rate": 3.474683544303798e-05,
      "loss": 0.194,
      "step": 1687
    },
    {
      "epoch": 2.14,
      "learning_rate": 3.4737793851717906e-05,
      "loss": 0.0468,
      "step": 1688
    },
    {
      "epoch": 2.14,
      "learning_rate": 3.472875226039784e-05,
      "loss": 0.0848,
      "step": 1689
    },
    {
      "epoch": 2.14,
      "learning_rate": 3.471971066907776e-05,
      "loss": 0.1561,
      "step": 1690
    },
    {
      "epoch": 2.14,
      "learning_rate": 3.471066907775769e-05,
      "loss": 0.18,
      "step": 1691
    },
    {
      "epoch": 2.14,
      "learning_rate": 3.470162748643762e-05,
      "loss": 0.18,
      "step": 1692
    },
    {
      "epoch": 2.14,
      "learning_rate": 3.4692585895117543e-05,
      "loss": 0.0287,
      "step": 1693
    },
    {
      "epoch": 2.14,
      "learning_rate": 3.468354430379747e-05,
      "loss": 0.089,
      "step": 1694
    },
    {
      "epoch": 2.15,
      "learning_rate": 3.46745027124774e-05,
      "loss": 0.1601,
      "step": 1695
    },
    {
      "epoch": 2.15,
      "learning_rate": 3.4665461121157325e-05,
      "loss": 0.0119,
      "step": 1696
    },
    {
      "epoch": 2.15,
      "learning_rate": 3.465641952983725e-05,
      "loss": 0.0163,
      "step": 1697
    },
    {
      "epoch": 2.15,
      "learning_rate": 3.464737793851718e-05,
      "loss": 0.1425,
      "step": 1698
    },
    {
      "epoch": 2.15,
      "learning_rate": 3.4638336347197106e-05,
      "loss": 0.1925,
      "step": 1699
    },
    {
      "epoch": 2.15,
      "learning_rate": 3.462929475587703e-05,
      "loss": 0.2072,
      "step": 1700
    },
    {
      "epoch": 2.15,
      "learning_rate": 3.462025316455696e-05,
      "loss": 0.1784,
      "step": 1701
    },
    {
      "epoch": 2.15,
      "learning_rate": 3.461121157323689e-05,
      "loss": 0.2794,
      "step": 1702
    },
    {
      "epoch": 2.16,
      "learning_rate": 3.460216998191682e-05,
      "loss": 0.0862,
      "step": 1703
    },
    {
      "epoch": 2.16,
      "learning_rate": 3.4593128390596744e-05,
      "loss": 0.093,
      "step": 1704
    },
    {
      "epoch": 2.16,
      "learning_rate": 3.4584086799276676e-05,
      "loss": 0.1324,
      "step": 1705
    },
    {
      "epoch": 2.16,
      "learning_rate": 3.45750452079566e-05,
      "loss": 0.0298,
      "step": 1706
    },
    {
      "epoch": 2.16,
      "learning_rate": 3.4566003616636525e-05,
      "loss": 0.0885,
      "step": 1707
    },
    {
      "epoch": 2.16,
      "learning_rate": 3.455696202531646e-05,
      "loss": 0.055,
      "step": 1708
    },
    {
      "epoch": 2.16,
      "learning_rate": 3.454792043399638e-05,
      "loss": 0.0629,
      "step": 1709
    },
    {
      "epoch": 2.16,
      "learning_rate": 3.453887884267631e-05,
      "loss": 0.1495,
      "step": 1710
    },
    {
      "epoch": 2.17,
      "learning_rate": 3.452983725135624e-05,
      "loss": 0.1702,
      "step": 1711
    },
    {
      "epoch": 2.17,
      "learning_rate": 3.452079566003617e-05,
      "loss": 0.2194,
      "step": 1712
    },
    {
      "epoch": 2.17,
      "learning_rate": 3.4511754068716095e-05,
      "loss": 0.089,
      "step": 1713
    },
    {
      "epoch": 2.17,
      "learning_rate": 3.450271247739602e-05,
      "loss": 0.0592,
      "step": 1714
    },
    {
      "epoch": 2.17,
      "learning_rate": 3.449367088607595e-05,
      "loss": 0.1979,
      "step": 1715
    },
    {
      "epoch": 2.17,
      "learning_rate": 3.4484629294755876e-05,
      "loss": 0.4503,
      "step": 1716
    },
    {
      "epoch": 2.17,
      "learning_rate": 3.447558770343581e-05,
      "loss": 0.033,
      "step": 1717
    },
    {
      "epoch": 2.17,
      "learning_rate": 3.446654611211573e-05,
      "loss": 0.2214,
      "step": 1718
    },
    {
      "epoch": 2.18,
      "learning_rate": 3.4457504520795664e-05,
      "loss": 0.106,
      "step": 1719
    },
    {
      "epoch": 2.18,
      "learning_rate": 3.444846292947559e-05,
      "loss": 0.3438,
      "step": 1720
    },
    {
      "epoch": 2.18,
      "learning_rate": 3.443942133815552e-05,
      "loss": 0.209,
      "step": 1721
    },
    {
      "epoch": 2.18,
      "learning_rate": 3.4430379746835445e-05,
      "loss": 0.1238,
      "step": 1722
    },
    {
      "epoch": 2.18,
      "learning_rate": 3.442133815551537e-05,
      "loss": 0.0425,
      "step": 1723
    },
    {
      "epoch": 2.18,
      "learning_rate": 3.44122965641953e-05,
      "loss": 0.0641,
      "step": 1724
    },
    {
      "epoch": 2.18,
      "learning_rate": 3.440325497287523e-05,
      "loss": 0.1711,
      "step": 1725
    },
    {
      "epoch": 2.18,
      "learning_rate": 3.439421338155516e-05,
      "loss": 0.1225,
      "step": 1726
    },
    {
      "epoch": 2.19,
      "learning_rate": 3.438517179023508e-05,
      "loss": 0.0673,
      "step": 1727
    },
    {
      "epoch": 2.19,
      "learning_rate": 3.4376130198915015e-05,
      "loss": 0.0718,
      "step": 1728
    },
    {
      "epoch": 2.19,
      "learning_rate": 3.436708860759494e-05,
      "loss": 0.1467,
      "step": 1729
    },
    {
      "epoch": 2.19,
      "learning_rate": 3.4358047016274865e-05,
      "loss": 0.1004,
      "step": 1730
    },
    {
      "epoch": 2.19,
      "learning_rate": 3.4349005424954796e-05,
      "loss": 0.163,
      "step": 1731
    },
    {
      "epoch": 2.19,
      "learning_rate": 3.433996383363472e-05,
      "loss": 0.0714,
      "step": 1732
    },
    {
      "epoch": 2.19,
      "learning_rate": 3.433092224231465e-05,
      "loss": 0.1229,
      "step": 1733
    },
    {
      "epoch": 2.19,
      "learning_rate": 3.432188065099458e-05,
      "loss": 0.0201,
      "step": 1734
    },
    {
      "epoch": 2.2,
      "learning_rate": 3.431283905967451e-05,
      "loss": 0.0425,
      "step": 1735
    },
    {
      "epoch": 2.2,
      "learning_rate": 3.4303797468354434e-05,
      "loss": 0.1565,
      "step": 1736
    },
    {
      "epoch": 2.2,
      "learning_rate": 3.4294755877034366e-05,
      "loss": 0.3198,
      "step": 1737
    },
    {
      "epoch": 2.2,
      "learning_rate": 3.428571428571429e-05,
      "loss": 0.1612,
      "step": 1738
    },
    {
      "epoch": 2.2,
      "learning_rate": 3.4276672694394215e-05,
      "loss": 0.0599,
      "step": 1739
    },
    {
      "epoch": 2.2,
      "learning_rate": 3.426763110307415e-05,
      "loss": 0.0249,
      "step": 1740
    },
    {
      "epoch": 2.2,
      "learning_rate": 3.425858951175407e-05,
      "loss": 0.0783,
      "step": 1741
    },
    {
      "epoch": 2.2,
      "learning_rate": 3.4249547920434e-05,
      "loss": 0.1121,
      "step": 1742
    },
    {
      "epoch": 2.21,
      "learning_rate": 3.424050632911393e-05,
      "loss": 0.2617,
      "step": 1743
    },
    {
      "epoch": 2.21,
      "learning_rate": 3.423146473779385e-05,
      "loss": 0.0185,
      "step": 1744
    },
    {
      "epoch": 2.21,
      "learning_rate": 3.422242314647378e-05,
      "loss": 0.0304,
      "step": 1745
    },
    {
      "epoch": 2.21,
      "learning_rate": 3.421338155515371e-05,
      "loss": 0.061,
      "step": 1746
    },
    {
      "epoch": 2.21,
      "learning_rate": 3.4204339963833634e-05,
      "loss": 0.0447,
      "step": 1747
    },
    {
      "epoch": 2.21,
      "learning_rate": 3.419529837251356e-05,
      "loss": 0.0405,
      "step": 1748
    },
    {
      "epoch": 2.21,
      "learning_rate": 3.418625678119349e-05,
      "loss": 0.0494,
      "step": 1749
    },
    {
      "epoch": 2.22,
      "learning_rate": 3.4177215189873416e-05,
      "loss": 0.0893,
      "step": 1750
    },
    {
      "epoch": 2.22,
      "eval_accuracy": 0.9016516516516516,
      "eval_loss": 0.2877936065196991,
      "eval_runtime": 45.1137,
      "eval_samples_per_second": 29.525,
      "eval_steps_per_second": 3.702,
      "step": 1750
    },
    {
      "epoch": 2.22,
      "learning_rate": 3.416817359855335e-05,
      "loss": 0.1234,
      "step": 1751
    },
    {
      "epoch": 2.22,
      "learning_rate": 3.415913200723327e-05,
      "loss": 0.0987,
      "step": 1752
    },
    {
      "epoch": 2.22,
      "learning_rate": 3.4150090415913204e-05,
      "loss": 0.1103,
      "step": 1753
    },
    {
      "epoch": 2.22,
      "learning_rate": 3.414104882459313e-05,
      "loss": 0.2265,
      "step": 1754
    },
    {
      "epoch": 2.22,
      "learning_rate": 3.4132007233273054e-05,
      "loss": 0.0699,
      "step": 1755
    },
    {
      "epoch": 2.22,
      "learning_rate": 3.4122965641952985e-05,
      "loss": 0.0674,
      "step": 1756
    },
    {
      "epoch": 2.22,
      "learning_rate": 3.411392405063291e-05,
      "loss": 0.3132,
      "step": 1757
    },
    {
      "epoch": 2.23,
      "learning_rate": 3.410488245931284e-05,
      "loss": 0.1807,
      "step": 1758
    },
    {
      "epoch": 2.23,
      "learning_rate": 3.4095840867992767e-05,
      "loss": 0.1251,
      "step": 1759
    },
    {
      "epoch": 2.23,
      "learning_rate": 3.40867992766727e-05,
      "loss": 0.0981,
      "step": 1760
    },
    {
      "epoch": 2.23,
      "learning_rate": 3.407775768535262e-05,
      "loss": 0.0764,
      "step": 1761
    },
    {
      "epoch": 2.23,
      "learning_rate": 3.406871609403255e-05,
      "loss": 0.1564,
      "step": 1762
    },
    {
      "epoch": 2.23,
      "learning_rate": 3.405967450271248e-05,
      "loss": 0.0306,
      "step": 1763
    },
    {
      "epoch": 2.23,
      "learning_rate": 3.4050632911392404e-05,
      "loss": 0.0762,
      "step": 1764
    },
    {
      "epoch": 2.23,
      "learning_rate": 3.4041591320072336e-05,
      "loss": 0.1171,
      "step": 1765
    },
    {
      "epoch": 2.24,
      "learning_rate": 3.403254972875226e-05,
      "loss": 0.0586,
      "step": 1766
    },
    {
      "epoch": 2.24,
      "learning_rate": 3.402350813743219e-05,
      "loss": 0.1353,
      "step": 1767
    },
    {
      "epoch": 2.24,
      "learning_rate": 3.401446654611212e-05,
      "loss": 0.0739,
      "step": 1768
    },
    {
      "epoch": 2.24,
      "learning_rate": 3.400542495479204e-05,
      "loss": 0.1174,
      "step": 1769
    },
    {
      "epoch": 2.24,
      "learning_rate": 3.3996383363471974e-05,
      "loss": 0.2216,
      "step": 1770
    },
    {
      "epoch": 2.24,
      "learning_rate": 3.39873417721519e-05,
      "loss": 0.0831,
      "step": 1771
    },
    {
      "epoch": 2.24,
      "learning_rate": 3.397830018083183e-05,
      "loss": 0.0909,
      "step": 1772
    },
    {
      "epoch": 2.24,
      "learning_rate": 3.3969258589511755e-05,
      "loss": 0.1416,
      "step": 1773
    },
    {
      "epoch": 2.25,
      "learning_rate": 3.396021699819169e-05,
      "loss": 0.0118,
      "step": 1774
    },
    {
      "epoch": 2.25,
      "learning_rate": 3.395117540687161e-05,
      "loss": 0.0475,
      "step": 1775
    },
    {
      "epoch": 2.25,
      "learning_rate": 3.394213381555154e-05,
      "loss": 0.2108,
      "step": 1776
    },
    {
      "epoch": 2.25,
      "learning_rate": 3.393309222423147e-05,
      "loss": 0.1948,
      "step": 1777
    },
    {
      "epoch": 2.25,
      "learning_rate": 3.392405063291139e-05,
      "loss": 0.0832,
      "step": 1778
    },
    {
      "epoch": 2.25,
      "learning_rate": 3.3915009041591324e-05,
      "loss": 0.0409,
      "step": 1779
    },
    {
      "epoch": 2.25,
      "learning_rate": 3.390596745027125e-05,
      "loss": 0.1701,
      "step": 1780
    },
    {
      "epoch": 2.25,
      "learning_rate": 3.389692585895118e-05,
      "loss": 0.0963,
      "step": 1781
    },
    {
      "epoch": 2.26,
      "learning_rate": 3.3887884267631106e-05,
      "loss": 0.0288,
      "step": 1782
    },
    {
      "epoch": 2.26,
      "learning_rate": 3.387884267631104e-05,
      "loss": 0.1648,
      "step": 1783
    },
    {
      "epoch": 2.26,
      "learning_rate": 3.386980108499096e-05,
      "loss": 0.0675,
      "step": 1784
    },
    {
      "epoch": 2.26,
      "learning_rate": 3.386075949367089e-05,
      "loss": 0.3179,
      "step": 1785
    },
    {
      "epoch": 2.26,
      "learning_rate": 3.385171790235082e-05,
      "loss": 0.0469,
      "step": 1786
    },
    {
      "epoch": 2.26,
      "learning_rate": 3.3842676311030744e-05,
      "loss": 0.134,
      "step": 1787
    },
    {
      "epoch": 2.26,
      "learning_rate": 3.383363471971067e-05,
      "loss": 0.1163,
      "step": 1788
    },
    {
      "epoch": 2.26,
      "learning_rate": 3.38245931283906e-05,
      "loss": 0.1114,
      "step": 1789
    },
    {
      "epoch": 2.27,
      "learning_rate": 3.3815551537070525e-05,
      "loss": 0.158,
      "step": 1790
    },
    {
      "epoch": 2.27,
      "learning_rate": 3.380650994575045e-05,
      "loss": 0.0726,
      "step": 1791
    },
    {
      "epoch": 2.27,
      "learning_rate": 3.379746835443038e-05,
      "loss": 0.3121,
      "step": 1792
    },
    {
      "epoch": 2.27,
      "learning_rate": 3.3788426763110306e-05,
      "loss": 0.0747,
      "step": 1793
    },
    {
      "epoch": 2.27,
      "learning_rate": 3.377938517179023e-05,
      "loss": 0.0434,
      "step": 1794
    },
    {
      "epoch": 2.27,
      "learning_rate": 3.377034358047016e-05,
      "loss": 0.1015,
      "step": 1795
    },
    {
      "epoch": 2.27,
      "learning_rate": 3.376130198915009e-05,
      "loss": 0.108,
      "step": 1796
    },
    {
      "epoch": 2.27,
      "learning_rate": 3.375226039783002e-05,
      "loss": 0.0598,
      "step": 1797
    },
    {
      "epoch": 2.28,
      "learning_rate": 3.3743218806509944e-05,
      "loss": 0.0161,
      "step": 1798
    },
    {
      "epoch": 2.28,
      "learning_rate": 3.3734177215189876e-05,
      "loss": 0.0756,
      "step": 1799
    },
    {
      "epoch": 2.28,
      "learning_rate": 3.37251356238698e-05,
      "loss": 0.2761,
      "step": 1800
    },
    {
      "epoch": 2.28,
      "learning_rate": 3.3716094032549725e-05,
      "loss": 0.0133,
      "step": 1801
    },
    {
      "epoch": 2.28,
      "learning_rate": 3.370705244122966e-05,
      "loss": 0.0126,
      "step": 1802
    },
    {
      "epoch": 2.28,
      "learning_rate": 3.369801084990958e-05,
      "loss": 0.0338,
      "step": 1803
    },
    {
      "epoch": 2.28,
      "learning_rate": 3.3688969258589513e-05,
      "loss": 0.1215,
      "step": 1804
    },
    {
      "epoch": 2.28,
      "learning_rate": 3.367992766726944e-05,
      "loss": 0.032,
      "step": 1805
    },
    {
      "epoch": 2.29,
      "learning_rate": 3.367088607594937e-05,
      "loss": 0.0364,
      "step": 1806
    },
    {
      "epoch": 2.29,
      "learning_rate": 3.3661844484629295e-05,
      "loss": 0.0251,
      "step": 1807
    },
    {
      "epoch": 2.29,
      "learning_rate": 3.3652802893309226e-05,
      "loss": 0.1088,
      "step": 1808
    },
    {
      "epoch": 2.29,
      "learning_rate": 3.364376130198915e-05,
      "loss": 0.2437,
      "step": 1809
    },
    {
      "epoch": 2.29,
      "learning_rate": 3.3634719710669076e-05,
      "loss": 0.3386,
      "step": 1810
    },
    {
      "epoch": 2.29,
      "learning_rate": 3.362567811934901e-05,
      "loss": 0.2092,
      "step": 1811
    },
    {
      "epoch": 2.29,
      "learning_rate": 3.361663652802893e-05,
      "loss": 0.0484,
      "step": 1812
    },
    {
      "epoch": 2.29,
      "learning_rate": 3.3607594936708864e-05,
      "loss": 0.0749,
      "step": 1813
    },
    {
      "epoch": 2.3,
      "learning_rate": 3.359855334538879e-05,
      "loss": 0.1855,
      "step": 1814
    },
    {
      "epoch": 2.3,
      "learning_rate": 3.358951175406872e-05,
      "loss": 0.1122,
      "step": 1815
    },
    {
      "epoch": 2.3,
      "learning_rate": 3.3580470162748646e-05,
      "loss": 0.0892,
      "step": 1816
    },
    {
      "epoch": 2.3,
      "learning_rate": 3.357142857142857e-05,
      "loss": 0.0286,
      "step": 1817
    },
    {
      "epoch": 2.3,
      "learning_rate": 3.35623869801085e-05,
      "loss": 0.122,
      "step": 1818
    },
    {
      "epoch": 2.3,
      "learning_rate": 3.355334538878843e-05,
      "loss": 0.0756,
      "step": 1819
    },
    {
      "epoch": 2.3,
      "learning_rate": 3.354430379746836e-05,
      "loss": 0.245,
      "step": 1820
    },
    {
      "epoch": 2.3,
      "learning_rate": 3.353526220614828e-05,
      "loss": 0.1339,
      "step": 1821
    },
    {
      "epoch": 2.31,
      "learning_rate": 3.3526220614828215e-05,
      "loss": 0.112,
      "step": 1822
    },
    {
      "epoch": 2.31,
      "learning_rate": 3.351717902350814e-05,
      "loss": 0.0804,
      "step": 1823
    },
    {
      "epoch": 2.31,
      "learning_rate": 3.350813743218807e-05,
      "loss": 0.0568,
      "step": 1824
    },
    {
      "epoch": 2.31,
      "learning_rate": 3.3499095840867996e-05,
      "loss": 0.0953,
      "step": 1825
    },
    {
      "epoch": 2.31,
      "learning_rate": 3.349005424954792e-05,
      "loss": 0.1078,
      "step": 1826
    },
    {
      "epoch": 2.31,
      "learning_rate": 3.348101265822785e-05,
      "loss": 0.1208,
      "step": 1827
    },
    {
      "epoch": 2.31,
      "learning_rate": 3.347197106690778e-05,
      "loss": 0.1582,
      "step": 1828
    },
    {
      "epoch": 2.31,
      "learning_rate": 3.346292947558771e-05,
      "loss": 0.0913,
      "step": 1829
    },
    {
      "epoch": 2.32,
      "learning_rate": 3.3453887884267634e-05,
      "loss": 0.2216,
      "step": 1830
    },
    {
      "epoch": 2.32,
      "learning_rate": 3.3444846292947566e-05,
      "loss": 0.1086,
      "step": 1831
    },
    {
      "epoch": 2.32,
      "learning_rate": 3.343580470162749e-05,
      "loss": 0.0635,
      "step": 1832
    },
    {
      "epoch": 2.32,
      "learning_rate": 3.3426763110307415e-05,
      "loss": 0.0849,
      "step": 1833
    },
    {
      "epoch": 2.32,
      "learning_rate": 3.341772151898735e-05,
      "loss": 0.1077,
      "step": 1834
    },
    {
      "epoch": 2.32,
      "learning_rate": 3.340867992766727e-05,
      "loss": 0.0574,
      "step": 1835
    },
    {
      "epoch": 2.32,
      "learning_rate": 3.33996383363472e-05,
      "loss": 0.0384,
      "step": 1836
    },
    {
      "epoch": 2.33,
      "learning_rate": 3.339059674502713e-05,
      "loss": 0.0509,
      "step": 1837
    },
    {
      "epoch": 2.33,
      "learning_rate": 3.338155515370705e-05,
      "loss": 0.1189,
      "step": 1838
    },
    {
      "epoch": 2.33,
      "learning_rate": 3.337251356238698e-05,
      "loss": 0.1147,
      "step": 1839
    },
    {
      "epoch": 2.33,
      "learning_rate": 3.336347197106691e-05,
      "loss": 0.0529,
      "step": 1840
    },
    {
      "epoch": 2.33,
      "learning_rate": 3.3354430379746835e-05,
      "loss": 0.0665,
      "step": 1841
    },
    {
      "epoch": 2.33,
      "learning_rate": 3.334538878842676e-05,
      "loss": 0.0163,
      "step": 1842
    },
    {
      "epoch": 2.33,
      "learning_rate": 3.333634719710669e-05,
      "loss": 0.3462,
      "step": 1843
    },
    {
      "epoch": 2.33,
      "learning_rate": 3.3327305605786616e-05,
      "loss": 0.0866,
      "step": 1844
    },
    {
      "epoch": 2.34,
      "learning_rate": 3.331826401446655e-05,
      "loss": 0.0327,
      "step": 1845
    },
    {
      "epoch": 2.34,
      "learning_rate": 3.330922242314647e-05,
      "loss": 0.2658,
      "step": 1846
    },
    {
      "epoch": 2.34,
      "learning_rate": 3.3300180831826404e-05,
      "loss": 0.0497,
      "step": 1847
    },
    {
      "epoch": 2.34,
      "learning_rate": 3.329113924050633e-05,
      "loss": 0.2795,
      "step": 1848
    },
    {
      "epoch": 2.34,
      "learning_rate": 3.3282097649186254e-05,
      "loss": 0.1165,
      "step": 1849
    },
    {
      "epoch": 2.34,
      "learning_rate": 3.3273056057866185e-05,
      "loss": 0.0777,
      "step": 1850
    },
    {
      "epoch": 2.34,
      "learning_rate": 3.326401446654611e-05,
      "loss": 0.0575,
      "step": 1851
    },
    {
      "epoch": 2.34,
      "learning_rate": 3.325497287522604e-05,
      "loss": 0.0765,
      "step": 1852
    },
    {
      "epoch": 2.35,
      "learning_rate": 3.324593128390597e-05,
      "loss": 0.0703,
      "step": 1853
    },
    {
      "epoch": 2.35,
      "learning_rate": 3.32368896925859e-05,
      "loss": 0.046,
      "step": 1854
    },
    {
      "epoch": 2.35,
      "learning_rate": 3.322784810126582e-05,
      "loss": 0.4007,
      "step": 1855
    },
    {
      "epoch": 2.35,
      "learning_rate": 3.3218806509945755e-05,
      "loss": 0.1314,
      "step": 1856
    },
    {
      "epoch": 2.35,
      "learning_rate": 3.320976491862568e-05,
      "loss": 0.1814,
      "step": 1857
    },
    {
      "epoch": 2.35,
      "learning_rate": 3.3200723327305604e-05,
      "loss": 0.1291,
      "step": 1858
    },
    {
      "epoch": 2.35,
      "learning_rate": 3.3191681735985536e-05,
      "loss": 0.1323,
      "step": 1859
    },
    {
      "epoch": 2.35,
      "learning_rate": 3.318264014466546e-05,
      "loss": 0.1524,
      "step": 1860
    },
    {
      "epoch": 2.36,
      "learning_rate": 3.317359855334539e-05,
      "loss": 0.2073,
      "step": 1861
    },
    {
      "epoch": 2.36,
      "learning_rate": 3.316455696202532e-05,
      "loss": 0.0964,
      "step": 1862
    },
    {
      "epoch": 2.36,
      "learning_rate": 3.315551537070525e-05,
      "loss": 0.0592,
      "step": 1863
    },
    {
      "epoch": 2.36,
      "learning_rate": 3.3146473779385174e-05,
      "loss": 0.0734,
      "step": 1864
    },
    {
      "epoch": 2.36,
      "learning_rate": 3.31374321880651e-05,
      "loss": 0.2159,
      "step": 1865
    },
    {
      "epoch": 2.36,
      "learning_rate": 3.312839059674503e-05,
      "loss": 0.0255,
      "step": 1866
    },
    {
      "epoch": 2.36,
      "learning_rate": 3.3119349005424955e-05,
      "loss": 0.1332,
      "step": 1867
    },
    {
      "epoch": 2.36,
      "learning_rate": 3.311030741410489e-05,
      "loss": 0.18,
      "step": 1868
    },
    {
      "epoch": 2.37,
      "learning_rate": 3.310126582278481e-05,
      "loss": 0.2639,
      "step": 1869
    },
    {
      "epoch": 2.37,
      "learning_rate": 3.309222423146474e-05,
      "loss": 0.2625,
      "step": 1870
    },
    {
      "epoch": 2.37,
      "learning_rate": 3.308318264014467e-05,
      "loss": 0.0747,
      "step": 1871
    },
    {
      "epoch": 2.37,
      "learning_rate": 3.30741410488246e-05,
      "loss": 0.0436,
      "step": 1872
    },
    {
      "epoch": 2.37,
      "learning_rate": 3.3065099457504525e-05,
      "loss": 0.0718,
      "step": 1873
    },
    {
      "epoch": 2.37,
      "learning_rate": 3.305605786618445e-05,
      "loss": 0.1308,
      "step": 1874
    },
    {
      "epoch": 2.37,
      "learning_rate": 3.304701627486438e-05,
      "loss": 0.1286,
      "step": 1875
    },
    {
      "epoch": 2.37,
      "learning_rate": 3.3037974683544306e-05,
      "loss": 0.0425,
      "step": 1876
    },
    {
      "epoch": 2.38,
      "learning_rate": 3.302893309222424e-05,
      "loss": 0.0946,
      "step": 1877
    },
    {
      "epoch": 2.38,
      "learning_rate": 3.301989150090416e-05,
      "loss": 0.2662,
      "step": 1878
    },
    {
      "epoch": 2.38,
      "learning_rate": 3.301084990958409e-05,
      "loss": 0.1724,
      "step": 1879
    },
    {
      "epoch": 2.38,
      "learning_rate": 3.300180831826402e-05,
      "loss": 0.1195,
      "step": 1880
    },
    {
      "epoch": 2.38,
      "learning_rate": 3.2992766726943944e-05,
      "loss": 0.0397,
      "step": 1881
    },
    {
      "epoch": 2.38,
      "learning_rate": 3.298372513562387e-05,
      "loss": 0.07,
      "step": 1882
    },
    {
      "epoch": 2.38,
      "learning_rate": 3.29746835443038e-05,
      "loss": 0.0638,
      "step": 1883
    },
    {
      "epoch": 2.38,
      "learning_rate": 3.2965641952983725e-05,
      "loss": 0.1153,
      "step": 1884
    },
    {
      "epoch": 2.39,
      "learning_rate": 3.295660036166365e-05,
      "loss": 0.073,
      "step": 1885
    },
    {
      "epoch": 2.39,
      "learning_rate": 3.294755877034358e-05,
      "loss": 0.1495,
      "step": 1886
    },
    {
      "epoch": 2.39,
      "learning_rate": 3.2938517179023506e-05,
      "loss": 0.2607,
      "step": 1887
    },
    {
      "epoch": 2.39,
      "learning_rate": 3.292947558770344e-05,
      "loss": 0.1205,
      "step": 1888
    },
    {
      "epoch": 2.39,
      "learning_rate": 3.292043399638336e-05,
      "loss": 0.0448,
      "step": 1889
    },
    {
      "epoch": 2.39,
      "learning_rate": 3.291139240506329e-05,
      "loss": 0.1958,
      "step": 1890
    },
    {
      "epoch": 2.39,
      "learning_rate": 3.290235081374322e-05,
      "loss": 0.0848,
      "step": 1891
    },
    {
      "epoch": 2.39,
      "learning_rate": 3.2893309222423144e-05,
      "loss": 0.1716,
      "step": 1892
    },
    {
      "epoch": 2.4,
      "learning_rate": 3.2884267631103076e-05,
      "loss": 0.4517,
      "step": 1893
    },
    {
      "epoch": 2.4,
      "learning_rate": 3.2875226039783e-05,
      "loss": 0.065,
      "step": 1894
    },
    {
      "epoch": 2.4,
      "learning_rate": 3.286618444846293e-05,
      "loss": 0.1486,
      "step": 1895
    },
    {
      "epoch": 2.4,
      "learning_rate": 3.285714285714286e-05,
      "loss": 0.0571,
      "step": 1896
    },
    {
      "epoch": 2.4,
      "learning_rate": 3.284810126582278e-05,
      "loss": 0.1232,
      "step": 1897
    },
    {
      "epoch": 2.4,
      "learning_rate": 3.2839059674502714e-05,
      "loss": 0.0583,
      "step": 1898
    },
    {
      "epoch": 2.4,
      "learning_rate": 3.283001808318264e-05,
      "loss": 0.2478,
      "step": 1899
    },
    {
      "epoch": 2.4,
      "learning_rate": 3.282097649186257e-05,
      "loss": 0.0322,
      "step": 1900
    },
    {
      "epoch": 2.41,
      "learning_rate": 3.2811934900542495e-05,
      "loss": 0.0897,
      "step": 1901
    },
    {
      "epoch": 2.41,
      "learning_rate": 3.2802893309222427e-05,
      "loss": 0.1406,
      "step": 1902
    },
    {
      "epoch": 2.41,
      "learning_rate": 3.279385171790235e-05,
      "loss": 0.1275,
      "step": 1903
    },
    {
      "epoch": 2.41,
      "learning_rate": 3.278481012658228e-05,
      "loss": 0.0973,
      "step": 1904
    },
    {
      "epoch": 2.41,
      "learning_rate": 3.277576853526221e-05,
      "loss": 0.0156,
      "step": 1905
    },
    {
      "epoch": 2.41,
      "learning_rate": 3.276672694394213e-05,
      "loss": 0.1861,
      "step": 1906
    },
    {
      "epoch": 2.41,
      "learning_rate": 3.2757685352622064e-05,
      "loss": 0.0762,
      "step": 1907
    },
    {
      "epoch": 2.41,
      "learning_rate": 3.274864376130199e-05,
      "loss": 0.0787,
      "step": 1908
    },
    {
      "epoch": 2.42,
      "learning_rate": 3.273960216998192e-05,
      "loss": 0.0383,
      "step": 1909
    },
    {
      "epoch": 2.42,
      "learning_rate": 3.2730560578661846e-05,
      "loss": 0.1477,
      "step": 1910
    },
    {
      "epoch": 2.42,
      "learning_rate": 3.272151898734178e-05,
      "loss": 0.2854,
      "step": 1911
    },
    {
      "epoch": 2.42,
      "learning_rate": 3.27124773960217e-05,
      "loss": 0.0764,
      "step": 1912
    },
    {
      "epoch": 2.42,
      "learning_rate": 3.270343580470163e-05,
      "loss": 0.0785,
      "step": 1913
    },
    {
      "epoch": 2.42,
      "learning_rate": 3.269439421338156e-05,
      "loss": 0.234,
      "step": 1914
    },
    {
      "epoch": 2.42,
      "learning_rate": 3.2685352622061484e-05,
      "loss": 0.1051,
      "step": 1915
    },
    {
      "epoch": 2.43,
      "learning_rate": 3.2676311030741415e-05,
      "loss": 0.1471,
      "step": 1916
    },
    {
      "epoch": 2.43,
      "learning_rate": 3.266726943942134e-05,
      "loss": 0.275,
      "step": 1917
    },
    {
      "epoch": 2.43,
      "learning_rate": 3.265822784810127e-05,
      "loss": 0.0691,
      "step": 1918
    },
    {
      "epoch": 2.43,
      "learning_rate": 3.2649186256781196e-05,
      "loss": 0.3197,
      "step": 1919
    },
    {
      "epoch": 2.43,
      "learning_rate": 3.264014466546113e-05,
      "loss": 0.1302,
      "step": 1920
    },
    {
      "epoch": 2.43,
      "learning_rate": 3.263110307414105e-05,
      "loss": 0.2881,
      "step": 1921
    },
    {
      "epoch": 2.43,
      "learning_rate": 3.262206148282098e-05,
      "loss": 0.1103,
      "step": 1922
    },
    {
      "epoch": 2.43,
      "learning_rate": 3.261301989150091e-05,
      "loss": 0.0164,
      "step": 1923
    },
    {
      "epoch": 2.44,
      "learning_rate": 3.2603978300180834e-05,
      "loss": 0.2595,
      "step": 1924
    },
    {
      "epoch": 2.44,
      "learning_rate": 3.2594936708860766e-05,
      "loss": 0.0289,
      "step": 1925
    },
    {
      "epoch": 2.44,
      "learning_rate": 3.258589511754069e-05,
      "loss": 0.1498,
      "step": 1926
    },
    {
      "epoch": 2.44,
      "learning_rate": 3.2576853526220616e-05,
      "loss": 0.0908,
      "step": 1927
    },
    {
      "epoch": 2.44,
      "learning_rate": 3.256781193490055e-05,
      "loss": 0.0394,
      "step": 1928
    },
    {
      "epoch": 2.44,
      "learning_rate": 3.255877034358047e-05,
      "loss": 0.0651,
      "step": 1929
    },
    {
      "epoch": 2.44,
      "learning_rate": 3.25497287522604e-05,
      "loss": 0.1356,
      "step": 1930
    },
    {
      "epoch": 2.44,
      "learning_rate": 3.254068716094033e-05,
      "loss": 0.1248,
      "step": 1931
    },
    {
      "epoch": 2.45,
      "learning_rate": 3.2531645569620253e-05,
      "loss": 0.2526,
      "step": 1932
    },
    {
      "epoch": 2.45,
      "learning_rate": 3.252260397830018e-05,
      "loss": 0.3147,
      "step": 1933
    },
    {
      "epoch": 2.45,
      "learning_rate": 3.251356238698011e-05,
      "loss": 0.192,
      "step": 1934
    },
    {
      "epoch": 2.45,
      "learning_rate": 3.2504520795660035e-05,
      "loss": 0.206,
      "step": 1935
    },
    {
      "epoch": 2.45,
      "learning_rate": 3.2495479204339966e-05,
      "loss": 0.1595,
      "step": 1936
    },
    {
      "epoch": 2.45,
      "learning_rate": 3.248643761301989e-05,
      "loss": 0.176,
      "step": 1937
    },
    {
      "epoch": 2.45,
      "learning_rate": 3.2477396021699816e-05,
      "loss": 0.1803,
      "step": 1938
    },
    {
      "epoch": 2.45,
      "learning_rate": 3.246835443037975e-05,
      "loss": 0.0418,
      "step": 1939
    },
    {
      "epoch": 2.46,
      "learning_rate": 3.245931283905967e-05,
      "loss": 0.0958,
      "step": 1940
    },
    {
      "epoch": 2.46,
      "learning_rate": 3.2450271247739604e-05,
      "loss": 0.0681,
      "step": 1941
    },
    {
      "epoch": 2.46,
      "learning_rate": 3.244122965641953e-05,
      "loss": 0.2604,
      "step": 1942
    },
    {
      "epoch": 2.46,
      "learning_rate": 3.243218806509946e-05,
      "loss": 0.2412,
      "step": 1943
    },
    {
      "epoch": 2.46,
      "learning_rate": 3.2423146473779385e-05,
      "loss": 0.2124,
      "step": 1944
    },
    {
      "epoch": 2.46,
      "learning_rate": 3.241410488245931e-05,
      "loss": 0.1603,
      "step": 1945
    },
    {
      "epoch": 2.46,
      "learning_rate": 3.240506329113924e-05,
      "loss": 0.0452,
      "step": 1946
    },
    {
      "epoch": 2.46,
      "learning_rate": 3.239602169981917e-05,
      "loss": 0.0227,
      "step": 1947
    },
    {
      "epoch": 2.47,
      "learning_rate": 3.23869801084991e-05,
      "loss": 0.1537,
      "step": 1948
    },
    {
      "epoch": 2.47,
      "learning_rate": 3.237793851717902e-05,
      "loss": 0.0468,
      "step": 1949
    },
    {
      "epoch": 2.47,
      "learning_rate": 3.2368896925858955e-05,
      "loss": 0.1906,
      "step": 1950
    },
    {
      "epoch": 2.47,
      "learning_rate": 3.235985533453888e-05,
      "loss": 0.21,
      "step": 1951
    },
    {
      "epoch": 2.47,
      "learning_rate": 3.235081374321881e-05,
      "loss": 0.137,
      "step": 1952
    },
    {
      "epoch": 2.47,
      "learning_rate": 3.2341772151898736e-05,
      "loss": 0.2587,
      "step": 1953
    },
    {
      "epoch": 2.47,
      "learning_rate": 3.233273056057866e-05,
      "loss": 0.1233,
      "step": 1954
    },
    {
      "epoch": 2.47,
      "learning_rate": 3.232368896925859e-05,
      "loss": 0.0952,
      "step": 1955
    },
    {
      "epoch": 2.48,
      "learning_rate": 3.231464737793852e-05,
      "loss": 0.0641,
      "step": 1956
    },
    {
      "epoch": 2.48,
      "learning_rate": 3.230560578661845e-05,
      "loss": 0.1065,
      "step": 1957
    },
    {
      "epoch": 2.48,
      "learning_rate": 3.2296564195298374e-05,
      "loss": 0.0413,
      "step": 1958
    },
    {
      "epoch": 2.48,
      "learning_rate": 3.2287522603978306e-05,
      "loss": 0.2975,
      "step": 1959
    },
    {
      "epoch": 2.48,
      "learning_rate": 3.227848101265823e-05,
      "loss": 0.2122,
      "step": 1960
    },
    {
      "epoch": 2.48,
      "learning_rate": 3.2269439421338155e-05,
      "loss": 0.0264,
      "step": 1961
    },
    {
      "epoch": 2.48,
      "learning_rate": 3.226039783001809e-05,
      "loss": 0.1372,
      "step": 1962
    },
    {
      "epoch": 2.48,
      "learning_rate": 3.225135623869801e-05,
      "loss": 0.1272,
      "step": 1963
    },
    {
      "epoch": 2.49,
      "learning_rate": 3.2242314647377943e-05,
      "loss": 0.368,
      "step": 1964
    },
    {
      "epoch": 2.49,
      "learning_rate": 3.223327305605787e-05,
      "loss": 0.0346,
      "step": 1965
    },
    {
      "epoch": 2.49,
      "learning_rate": 3.22242314647378e-05,
      "loss": 0.02,
      "step": 1966
    },
    {
      "epoch": 2.49,
      "learning_rate": 3.2215189873417725e-05,
      "loss": 0.132,
      "step": 1967
    },
    {
      "epoch": 2.49,
      "learning_rate": 3.2206148282097656e-05,
      "loss": 0.0272,
      "step": 1968
    },
    {
      "epoch": 2.49,
      "learning_rate": 3.219710669077758e-05,
      "loss": 0.0772,
      "step": 1969
    },
    {
      "epoch": 2.49,
      "learning_rate": 3.2188065099457506e-05,
      "loss": 0.2433,
      "step": 1970
    },
    {
      "epoch": 2.49,
      "learning_rate": 3.217902350813744e-05,
      "loss": 0.0997,
      "step": 1971
    },
    {
      "epoch": 2.5,
      "learning_rate": 3.216998191681736e-05,
      "loss": 0.0925,
      "step": 1972
    },
    {
      "epoch": 2.5,
      "learning_rate": 3.216094032549729e-05,
      "loss": 0.1292,
      "step": 1973
    },
    {
      "epoch": 2.5,
      "learning_rate": 3.215189873417722e-05,
      "loss": 0.203,
      "step": 1974
    },
    {
      "epoch": 2.5,
      "learning_rate": 3.2142857142857144e-05,
      "loss": 0.1824,
      "step": 1975
    },
    {
      "epoch": 2.5,
      "learning_rate": 3.213381555153707e-05,
      "loss": 0.1471,
      "step": 1976
    },
    {
      "epoch": 2.5,
      "learning_rate": 3.2124773960217e-05,
      "loss": 0.0129,
      "step": 1977
    },
    {
      "epoch": 2.5,
      "learning_rate": 3.2115732368896925e-05,
      "loss": 0.1857,
      "step": 1978
    },
    {
      "epoch": 2.5,
      "learning_rate": 3.210669077757685e-05,
      "loss": 0.1543,
      "step": 1979
    },
    {
      "epoch": 2.51,
      "learning_rate": 3.209764918625678e-05,
      "loss": 0.1552,
      "step": 1980
    },
    {
      "epoch": 2.51,
      "learning_rate": 3.2088607594936707e-05,
      "loss": 0.1195,
      "step": 1981
    },
    {
      "epoch": 2.51,
      "learning_rate": 3.207956600361664e-05,
      "loss": 0.0595,
      "step": 1982
    },
    {
      "epoch": 2.51,
      "learning_rate": 3.207052441229656e-05,
      "loss": 0.2714,
      "step": 1983
    },
    {
      "epoch": 2.51,
      "learning_rate": 3.2061482820976495e-05,
      "loss": 0.1355,
      "step": 1984
    },
    {
      "epoch": 2.51,
      "learning_rate": 3.205244122965642e-05,
      "loss": 0.3053,
      "step": 1985
    },
    {
      "epoch": 2.51,
      "learning_rate": 3.2043399638336344e-05,
      "loss": 0.1122,
      "step": 1986
    },
    {
      "epoch": 2.51,
      "learning_rate": 3.2034358047016276e-05,
      "loss": 0.1125,
      "step": 1987
    },
    {
      "epoch": 2.52,
      "learning_rate": 3.20253164556962e-05,
      "loss": 0.1082,
      "step": 1988
    },
    {
      "epoch": 2.52,
      "learning_rate": 3.201627486437613e-05,
      "loss": 0.1776,
      "step": 1989
    },
    {
      "epoch": 2.52,
      "learning_rate": 3.200723327305606e-05,
      "loss": 0.0537,
      "step": 1990
    },
    {
      "epoch": 2.52,
      "learning_rate": 3.199819168173599e-05,
      "loss": 0.1026,
      "step": 1991
    },
    {
      "epoch": 2.52,
      "learning_rate": 3.1989150090415914e-05,
      "loss": 0.0573,
      "step": 1992
    },
    {
      "epoch": 2.52,
      "learning_rate": 3.198010849909584e-05,
      "loss": 0.0398,
      "step": 1993
    },
    {
      "epoch": 2.52,
      "learning_rate": 3.197106690777577e-05,
      "loss": 0.0741,
      "step": 1994
    },
    {
      "epoch": 2.52,
      "learning_rate": 3.1962025316455695e-05,
      "loss": 0.2361,
      "step": 1995
    },
    {
      "epoch": 2.53,
      "learning_rate": 3.195298372513563e-05,
      "loss": 0.0551,
      "step": 1996
    },
    {
      "epoch": 2.53,
      "learning_rate": 3.194394213381555e-05,
      "loss": 0.1086,
      "step": 1997
    },
    {
      "epoch": 2.53,
      "learning_rate": 3.193490054249548e-05,
      "loss": 0.0307,
      "step": 1998
    },
    {
      "epoch": 2.53,
      "learning_rate": 3.192585895117541e-05,
      "loss": 0.1982,
      "step": 1999
    },
    {
      "epoch": 2.53,
      "learning_rate": 3.191681735985534e-05,
      "loss": 0.1422,
      "step": 2000
    },
    {
      "epoch": 2.53,
      "eval_accuracy": 0.9099099099099099,
      "eval_loss": 0.2822848856449127,
      "eval_runtime": 45.274,
      "eval_samples_per_second": 29.421,
      "eval_steps_per_second": 3.689,
      "step": 2000
    },
    {
      "epoch": 2.53,
      "learning_rate": 3.1907775768535265e-05,
      "loss": 0.2617,
      "step": 2001
    },
    {
      "epoch": 2.53,
      "learning_rate": 3.189873417721519e-05,
      "loss": 0.0221,
      "step": 2002
    },
    {
      "epoch": 2.54,
      "learning_rate": 3.188969258589512e-05,
      "loss": 0.0335,
      "step": 2003
    },
    {
      "epoch": 2.54,
      "learning_rate": 3.1880650994575046e-05,
      "loss": 0.1739,
      "step": 2004
    },
    {
      "epoch": 2.54,
      "learning_rate": 3.187160940325498e-05,
      "loss": 0.1255,
      "step": 2005
    },
    {
      "epoch": 2.54,
      "learning_rate": 3.18625678119349e-05,
      "loss": 0.0513,
      "step": 2006
    },
    {
      "epoch": 2.54,
      "learning_rate": 3.1853526220614834e-05,
      "loss": 0.1874,
      "step": 2007
    },
    {
      "epoch": 2.54,
      "learning_rate": 3.184448462929476e-05,
      "loss": 0.1175,
      "step": 2008
    },
    {
      "epoch": 2.54,
      "learning_rate": 3.1835443037974684e-05,
      "loss": 0.027,
      "step": 2009
    },
    {
      "epoch": 2.54,
      "learning_rate": 3.1826401446654615e-05,
      "loss": 0.1124,
      "step": 2010
    },
    {
      "epoch": 2.55,
      "learning_rate": 3.181735985533454e-05,
      "loss": 0.1203,
      "step": 2011
    },
    {
      "epoch": 2.55,
      "learning_rate": 3.180831826401447e-05,
      "loss": 0.0484,
      "step": 2012
    },
    {
      "epoch": 2.55,
      "learning_rate": 3.17992766726944e-05,
      "loss": 0.2502,
      "step": 2013
    },
    {
      "epoch": 2.55,
      "learning_rate": 3.179023508137433e-05,
      "loss": 0.1133,
      "step": 2014
    },
    {
      "epoch": 2.55,
      "learning_rate": 3.178119349005425e-05,
      "loss": 0.1214,
      "step": 2015
    },
    {
      "epoch": 2.55,
      "learning_rate": 3.177215189873418e-05,
      "loss": 0.0624,
      "step": 2016
    },
    {
      "epoch": 2.55,
      "learning_rate": 3.176311030741411e-05,
      "loss": 0.1724,
      "step": 2017
    },
    {
      "epoch": 2.55,
      "learning_rate": 3.1754068716094034e-05,
      "loss": 0.0317,
      "step": 2018
    },
    {
      "epoch": 2.56,
      "learning_rate": 3.1745027124773966e-05,
      "loss": 0.0373,
      "step": 2019
    },
    {
      "epoch": 2.56,
      "learning_rate": 3.173598553345389e-05,
      "loss": 0.0488,
      "step": 2020
    },
    {
      "epoch": 2.56,
      "learning_rate": 3.1726943942133816e-05,
      "loss": 0.0305,
      "step": 2021
    },
    {
      "epoch": 2.56,
      "learning_rate": 3.171790235081375e-05,
      "loss": 0.0559,
      "step": 2022
    },
    {
      "epoch": 2.56,
      "learning_rate": 3.170886075949367e-05,
      "loss": 0.1355,
      "step": 2023
    },
    {
      "epoch": 2.56,
      "learning_rate": 3.16998191681736e-05,
      "loss": 0.1388,
      "step": 2024
    },
    {
      "epoch": 2.56,
      "learning_rate": 3.169077757685353e-05,
      "loss": 0.1517,
      "step": 2025
    },
    {
      "epoch": 2.56,
      "learning_rate": 3.1681735985533454e-05,
      "loss": 0.068,
      "step": 2026
    },
    {
      "epoch": 2.57,
      "learning_rate": 3.167269439421338e-05,
      "loss": 0.0645,
      "step": 2027
    },
    {
      "epoch": 2.57,
      "learning_rate": 3.166365280289331e-05,
      "loss": 0.0696,
      "step": 2028
    },
    {
      "epoch": 2.57,
      "learning_rate": 3.1654611211573235e-05,
      "loss": 0.1608,
      "step": 2029
    },
    {
      "epoch": 2.57,
      "learning_rate": 3.1645569620253167e-05,
      "loss": 0.1574,
      "step": 2030
    },
    {
      "epoch": 2.57,
      "learning_rate": 3.163652802893309e-05,
      "loss": 0.0636,
      "step": 2031
    },
    {
      "epoch": 2.57,
      "learning_rate": 3.1627486437613016e-05,
      "loss": 0.3148,
      "step": 2032
    },
    {
      "epoch": 2.57,
      "learning_rate": 3.161844484629295e-05,
      "loss": 0.1162,
      "step": 2033
    },
    {
      "epoch": 2.57,
      "learning_rate": 3.160940325497287e-05,
      "loss": 0.2521,
      "step": 2034
    },
    {
      "epoch": 2.58,
      "learning_rate": 3.1600361663652804e-05,
      "loss": 0.0836,
      "step": 2035
    },
    {
      "epoch": 2.58,
      "learning_rate": 3.159132007233273e-05,
      "loss": 0.1985,
      "step": 2036
    },
    {
      "epoch": 2.58,
      "learning_rate": 3.158227848101266e-05,
      "loss": 0.1533,
      "step": 2037
    },
    {
      "epoch": 2.58,
      "learning_rate": 3.1573236889692586e-05,
      "loss": 0.1693,
      "step": 2038
    },
    {
      "epoch": 2.58,
      "learning_rate": 3.156419529837252e-05,
      "loss": 0.0273,
      "step": 2039
    },
    {
      "epoch": 2.58,
      "learning_rate": 3.155515370705244e-05,
      "loss": 0.1327,
      "step": 2040
    },
    {
      "epoch": 2.58,
      "learning_rate": 3.154611211573237e-05,
      "loss": 0.1927,
      "step": 2041
    },
    {
      "epoch": 2.58,
      "learning_rate": 3.15370705244123e-05,
      "loss": 0.073,
      "step": 2042
    },
    {
      "epoch": 2.59,
      "learning_rate": 3.1528028933092223e-05,
      "loss": 0.0323,
      "step": 2043
    },
    {
      "epoch": 2.59,
      "learning_rate": 3.1518987341772155e-05,
      "loss": 0.0231,
      "step": 2044
    },
    {
      "epoch": 2.59,
      "learning_rate": 3.150994575045208e-05,
      "loss": 0.0854,
      "step": 2045
    },
    {
      "epoch": 2.59,
      "learning_rate": 3.150090415913201e-05,
      "loss": 0.163,
      "step": 2046
    },
    {
      "epoch": 2.59,
      "learning_rate": 3.1491862567811936e-05,
      "loss": 0.0346,
      "step": 2047
    },
    {
      "epoch": 2.59,
      "learning_rate": 3.148282097649186e-05,
      "loss": 0.1832,
      "step": 2048
    },
    {
      "epoch": 2.59,
      "learning_rate": 3.147377938517179e-05,
      "loss": 0.0269,
      "step": 2049
    },
    {
      "epoch": 2.59,
      "learning_rate": 3.146473779385172e-05,
      "loss": 0.1976,
      "step": 2050
    },
    {
      "epoch": 2.6,
      "learning_rate": 3.145569620253165e-05,
      "loss": 0.0872,
      "step": 2051
    },
    {
      "epoch": 2.6,
      "learning_rate": 3.1446654611211574e-05,
      "loss": 0.1826,
      "step": 2052
    },
    {
      "epoch": 2.6,
      "learning_rate": 3.1437613019891506e-05,
      "loss": 0.0311,
      "step": 2053
    },
    {
      "epoch": 2.6,
      "learning_rate": 3.142857142857143e-05,
      "loss": 0.2463,
      "step": 2054
    },
    {
      "epoch": 2.6,
      "learning_rate": 3.141952983725136e-05,
      "loss": 0.0723,
      "step": 2055
    },
    {
      "epoch": 2.6,
      "learning_rate": 3.141048824593129e-05,
      "loss": 0.1485,
      "step": 2056
    },
    {
      "epoch": 2.6,
      "learning_rate": 3.140144665461121e-05,
      "loss": 0.1862,
      "step": 2057
    },
    {
      "epoch": 2.6,
      "learning_rate": 3.1392405063291144e-05,
      "loss": 0.009,
      "step": 2058
    },
    {
      "epoch": 2.61,
      "learning_rate": 3.138336347197107e-05,
      "loss": 0.0405,
      "step": 2059
    },
    {
      "epoch": 2.61,
      "learning_rate": 3.1374321880651e-05,
      "loss": 0.0907,
      "step": 2060
    },
    {
      "epoch": 2.61,
      "learning_rate": 3.1365280289330925e-05,
      "loss": 0.1337,
      "step": 2061
    },
    {
      "epoch": 2.61,
      "learning_rate": 3.1356238698010857e-05,
      "loss": 0.0961,
      "step": 2062
    },
    {
      "epoch": 2.61,
      "learning_rate": 3.134719710669078e-05,
      "loss": 0.1443,
      "step": 2063
    },
    {
      "epoch": 2.61,
      "learning_rate": 3.1338155515370706e-05,
      "loss": 0.1347,
      "step": 2064
    },
    {
      "epoch": 2.61,
      "learning_rate": 3.132911392405064e-05,
      "loss": 0.1665,
      "step": 2065
    },
    {
      "epoch": 2.61,
      "learning_rate": 3.132007233273056e-05,
      "loss": 0.1021,
      "step": 2066
    },
    {
      "epoch": 2.62,
      "learning_rate": 3.131103074141049e-05,
      "loss": 0.1478,
      "step": 2067
    },
    {
      "epoch": 2.62,
      "learning_rate": 3.130198915009042e-05,
      "loss": 0.1654,
      "step": 2068
    },
    {
      "epoch": 2.62,
      "learning_rate": 3.1292947558770344e-05,
      "loss": 0.201,
      "step": 2069
    },
    {
      "epoch": 2.62,
      "learning_rate": 3.128390596745027e-05,
      "loss": 0.2125,
      "step": 2070
    },
    {
      "epoch": 2.62,
      "learning_rate": 3.12748643761302e-05,
      "loss": 0.1094,
      "step": 2071
    },
    {
      "epoch": 2.62,
      "learning_rate": 3.1265822784810125e-05,
      "loss": 0.0174,
      "step": 2072
    },
    {
      "epoch": 2.62,
      "learning_rate": 3.125678119349005e-05,
      "loss": 0.389,
      "step": 2073
    },
    {
      "epoch": 2.62,
      "learning_rate": 3.124773960216998e-05,
      "loss": 0.0646,
      "step": 2074
    },
    {
      "epoch": 2.63,
      "learning_rate": 3.123869801084991e-05,
      "loss": 0.3273,
      "step": 2075
    },
    {
      "epoch": 2.63,
      "learning_rate": 3.122965641952984e-05,
      "loss": 0.1272,
      "step": 2076
    },
    {
      "epoch": 2.63,
      "learning_rate": 3.122061482820976e-05,
      "loss": 0.0702,
      "step": 2077
    },
    {
      "epoch": 2.63,
      "learning_rate": 3.1211573236889695e-05,
      "loss": 0.053,
      "step": 2078
    },
    {
      "epoch": 2.63,
      "learning_rate": 3.120253164556962e-05,
      "loss": 0.1456,
      "step": 2079
    },
    {
      "epoch": 2.63,
      "learning_rate": 3.1193490054249545e-05,
      "loss": 0.0652,
      "step": 2080
    },
    {
      "epoch": 2.63,
      "learning_rate": 3.1184448462929476e-05,
      "loss": 0.0675,
      "step": 2081
    },
    {
      "epoch": 2.64,
      "learning_rate": 3.11754068716094e-05,
      "loss": 0.1053,
      "step": 2082
    },
    {
      "epoch": 2.64,
      "learning_rate": 3.116636528028933e-05,
      "loss": 0.1314,
      "step": 2083
    },
    {
      "epoch": 2.64,
      "learning_rate": 3.115732368896926e-05,
      "loss": 0.0157,
      "step": 2084
    },
    {
      "epoch": 2.64,
      "learning_rate": 3.114828209764919e-05,
      "loss": 0.0268,
      "step": 2085
    },
    {
      "epoch": 2.64,
      "learning_rate": 3.1139240506329114e-05,
      "loss": 0.5528,
      "step": 2086
    },
    {
      "epoch": 2.64,
      "learning_rate": 3.1130198915009046e-05,
      "loss": 0.1856,
      "step": 2087
    },
    {
      "epoch": 2.64,
      "learning_rate": 3.112115732368897e-05,
      "loss": 0.1587,
      "step": 2088
    },
    {
      "epoch": 2.64,
      "learning_rate": 3.1112115732368895e-05,
      "loss": 0.1527,
      "step": 2089
    },
    {
      "epoch": 2.65,
      "learning_rate": 3.110307414104883e-05,
      "loss": 0.0607,
      "step": 2090
    },
    {
      "epoch": 2.65,
      "learning_rate": 3.109403254972875e-05,
      "loss": 0.0471,
      "step": 2091
    },
    {
      "epoch": 2.65,
      "learning_rate": 3.108499095840868e-05,
      "loss": 0.047,
      "step": 2092
    },
    {
      "epoch": 2.65,
      "learning_rate": 3.107594936708861e-05,
      "loss": 0.0795,
      "step": 2093
    },
    {
      "epoch": 2.65,
      "learning_rate": 3.106690777576854e-05,
      "loss": 0.3562,
      "step": 2094
    },
    {
      "epoch": 2.65,
      "learning_rate": 3.1057866184448465e-05,
      "loss": 0.0119,
      "step": 2095
    },
    {
      "epoch": 2.65,
      "learning_rate": 3.104882459312839e-05,
      "loss": 0.1714,
      "step": 2096
    },
    {
      "epoch": 2.65,
      "learning_rate": 3.103978300180832e-05,
      "loss": 0.1319,
      "step": 2097
    },
    {
      "epoch": 2.66,
      "learning_rate": 3.1030741410488246e-05,
      "loss": 0.0782,
      "step": 2098
    },
    {
      "epoch": 2.66,
      "learning_rate": 3.102169981916818e-05,
      "loss": 0.2086,
      "step": 2099
    },
    {
      "epoch": 2.66,
      "learning_rate": 3.10126582278481e-05,
      "loss": 0.0614,
      "step": 2100
    },
    {
      "epoch": 2.66,
      "learning_rate": 3.1003616636528034e-05,
      "loss": 0.1,
      "step": 2101
    },
    {
      "epoch": 2.66,
      "learning_rate": 3.099457504520796e-05,
      "loss": 0.2098,
      "step": 2102
    },
    {
      "epoch": 2.66,
      "learning_rate": 3.098553345388789e-05,
      "loss": 0.2474,
      "step": 2103
    },
    {
      "epoch": 2.66,
      "learning_rate": 3.0976491862567815e-05,
      "loss": 0.1794,
      "step": 2104
    },
    {
      "epoch": 2.66,
      "learning_rate": 3.096745027124774e-05,
      "loss": 0.0799,
      "step": 2105
    },
    {
      "epoch": 2.67,
      "learning_rate": 3.095840867992767e-05,
      "loss": 0.0429,
      "step": 2106
    },
    {
      "epoch": 2.67,
      "learning_rate": 3.09493670886076e-05,
      "loss": 0.1409,
      "step": 2107
    },
    {
      "epoch": 2.67,
      "learning_rate": 3.094032549728753e-05,
      "loss": 0.2285,
      "step": 2108
    },
    {
      "epoch": 2.67,
      "learning_rate": 3.093128390596745e-05,
      "loss": 0.0111,
      "step": 2109
    },
    {
      "epoch": 2.67,
      "learning_rate": 3.0922242314647385e-05,
      "loss": 0.0234,
      "step": 2110
    },
    {
      "epoch": 2.67,
      "learning_rate": 3.091320072332731e-05,
      "loss": 0.1232,
      "step": 2111
    },
    {
      "epoch": 2.67,
      "learning_rate": 3.0904159132007235e-05,
      "loss": 0.3812,
      "step": 2112
    },
    {
      "epoch": 2.67,
      "learning_rate": 3.0895117540687166e-05,
      "loss": 0.0152,
      "step": 2113
    },
    {
      "epoch": 2.68,
      "learning_rate": 3.088607594936709e-05,
      "loss": 0.0602,
      "step": 2114
    },
    {
      "epoch": 2.68,
      "learning_rate": 3.0877034358047016e-05,
      "loss": 0.1072,
      "step": 2115
    },
    {
      "epoch": 2.68,
      "learning_rate": 3.086799276672695e-05,
      "loss": 0.0582,
      "step": 2116
    },
    {
      "epoch": 2.68,
      "learning_rate": 3.085895117540687e-05,
      "loss": 0.0681,
      "step": 2117
    },
    {
      "epoch": 2.68,
      "learning_rate": 3.08499095840868e-05,
      "loss": 0.0623,
      "step": 2118
    },
    {
      "epoch": 2.68,
      "learning_rate": 3.084086799276673e-05,
      "loss": 0.0454,
      "step": 2119
    },
    {
      "epoch": 2.68,
      "learning_rate": 3.0831826401446654e-05,
      "loss": 0.0628,
      "step": 2120
    },
    {
      "epoch": 2.68,
      "learning_rate": 3.082278481012658e-05,
      "loss": 0.0457,
      "step": 2121
    },
    {
      "epoch": 2.69,
      "learning_rate": 3.081374321880651e-05,
      "loss": 0.0209,
      "step": 2122
    },
    {
      "epoch": 2.69,
      "learning_rate": 3.0804701627486435e-05,
      "loss": 0.0449,
      "step": 2123
    },
    {
      "epoch": 2.69,
      "learning_rate": 3.079566003616637e-05,
      "loss": 0.1025,
      "step": 2124
    },
    {
      "epoch": 2.69,
      "learning_rate": 3.078661844484629e-05,
      "loss": 0.073,
      "step": 2125
    },
    {
      "epoch": 2.69,
      "learning_rate": 3.077757685352622e-05,
      "loss": 0.0779,
      "step": 2126
    },
    {
      "epoch": 2.69,
      "learning_rate": 3.076853526220615e-05,
      "loss": 0.0624,
      "step": 2127
    },
    {
      "epoch": 2.69,
      "learning_rate": 3.075949367088607e-05,
      "loss": 0.2202,
      "step": 2128
    },
    {
      "epoch": 2.69,
      "learning_rate": 3.0750452079566004e-05,
      "loss": 0.0229,
      "step": 2129
    },
    {
      "epoch": 2.7,
      "learning_rate": 3.074141048824593e-05,
      "loss": 0.0873,
      "step": 2130
    },
    {
      "epoch": 2.7,
      "learning_rate": 3.073236889692586e-05,
      "loss": 0.0571,
      "step": 2131
    },
    {
      "epoch": 2.7,
      "learning_rate": 3.0723327305605786e-05,
      "loss": 0.0887,
      "step": 2132
    },
    {
      "epoch": 2.7,
      "learning_rate": 3.071428571428572e-05,
      "loss": 0.0503,
      "step": 2133
    },
    {
      "epoch": 2.7,
      "learning_rate": 3.070524412296564e-05,
      "loss": 0.026,
      "step": 2134
    },
    {
      "epoch": 2.7,
      "learning_rate": 3.0696202531645574e-05,
      "loss": 0.1308,
      "step": 2135
    },
    {
      "epoch": 2.7,
      "learning_rate": 3.06871609403255e-05,
      "loss": 0.0778,
      "step": 2136
    },
    {
      "epoch": 2.7,
      "learning_rate": 3.0678119349005424e-05,
      "loss": 0.041,
      "step": 2137
    },
    {
      "epoch": 2.71,
      "learning_rate": 3.0669077757685355e-05,
      "loss": 0.0926,
      "step": 2138
    },
    {
      "epoch": 2.71,
      "learning_rate": 3.066003616636528e-05,
      "loss": 0.0105,
      "step": 2139
    },
    {
      "epoch": 2.71,
      "learning_rate": 3.065099457504521e-05,
      "loss": 0.1579,
      "step": 2140
    },
    {
      "epoch": 2.71,
      "learning_rate": 3.0641952983725137e-05,
      "loss": 0.245,
      "step": 2141
    },
    {
      "epoch": 2.71,
      "learning_rate": 3.063291139240507e-05,
      "loss": 0.0188,
      "step": 2142
    },
    {
      "epoch": 2.71,
      "learning_rate": 3.062386980108499e-05,
      "loss": 0.2217,
      "step": 2143
    },
    {
      "epoch": 2.71,
      "learning_rate": 3.061482820976492e-05,
      "loss": 0.1238,
      "step": 2144
    },
    {
      "epoch": 2.71,
      "learning_rate": 3.060578661844485e-05,
      "loss": 0.0532,
      "step": 2145
    },
    {
      "epoch": 2.72,
      "learning_rate": 3.0596745027124774e-05,
      "loss": 0.1649,
      "step": 2146
    },
    {
      "epoch": 2.72,
      "learning_rate": 3.0587703435804706e-05,
      "loss": 0.3624,
      "step": 2147
    },
    {
      "epoch": 2.72,
      "learning_rate": 3.057866184448463e-05,
      "loss": 0.0131,
      "step": 2148
    },
    {
      "epoch": 2.72,
      "learning_rate": 3.056962025316456e-05,
      "loss": 0.0739,
      "step": 2149
    },
    {
      "epoch": 2.72,
      "learning_rate": 3.056057866184449e-05,
      "loss": 0.0219,
      "step": 2150
    },
    {
      "epoch": 2.72,
      "learning_rate": 3.055153707052442e-05,
      "loss": 0.1382,
      "step": 2151
    },
    {
      "epoch": 2.72,
      "learning_rate": 3.0542495479204344e-05,
      "loss": 0.1056,
      "step": 2152
    },
    {
      "epoch": 2.72,
      "learning_rate": 3.053345388788427e-05,
      "loss": 0.1765,
      "step": 2153
    },
    {
      "epoch": 2.73,
      "learning_rate": 3.05244122965642e-05,
      "loss": 0.1351,
      "step": 2154
    },
    {
      "epoch": 2.73,
      "learning_rate": 3.0515370705244122e-05,
      "loss": 0.0851,
      "step": 2155
    },
    {
      "epoch": 2.73,
      "learning_rate": 3.0506329113924053e-05,
      "loss": 0.05,
      "step": 2156
    },
    {
      "epoch": 2.73,
      "learning_rate": 3.0497287522603978e-05,
      "loss": 0.27,
      "step": 2157
    },
    {
      "epoch": 2.73,
      "learning_rate": 3.048824593128391e-05,
      "loss": 0.0191,
      "step": 2158
    },
    {
      "epoch": 2.73,
      "learning_rate": 3.0479204339963835e-05,
      "loss": 0.1712,
      "step": 2159
    },
    {
      "epoch": 2.73,
      "learning_rate": 3.047016274864376e-05,
      "loss": 0.2556,
      "step": 2160
    },
    {
      "epoch": 2.73,
      "learning_rate": 3.046112115732369e-05,
      "loss": 0.0449,
      "step": 2161
    },
    {
      "epoch": 2.74,
      "learning_rate": 3.0452079566003616e-05,
      "loss": 0.0486,
      "step": 2162
    },
    {
      "epoch": 2.74,
      "learning_rate": 3.0443037974683548e-05,
      "loss": 0.0236,
      "step": 2163
    },
    {
      "epoch": 2.74,
      "learning_rate": 3.0433996383363472e-05,
      "loss": 0.0729,
      "step": 2164
    },
    {
      "epoch": 2.74,
      "learning_rate": 3.0424954792043404e-05,
      "loss": 0.1844,
      "step": 2165
    },
    {
      "epoch": 2.74,
      "learning_rate": 3.041591320072333e-05,
      "loss": 0.0222,
      "step": 2166
    },
    {
      "epoch": 2.74,
      "learning_rate": 3.040687160940326e-05,
      "loss": 0.4105,
      "step": 2167
    },
    {
      "epoch": 2.74,
      "learning_rate": 3.0397830018083185e-05,
      "loss": 0.0214,
      "step": 2168
    },
    {
      "epoch": 2.75,
      "learning_rate": 3.038878842676311e-05,
      "loss": 0.355,
      "step": 2169
    },
    {
      "epoch": 2.75,
      "learning_rate": 3.0379746835443042e-05,
      "loss": 0.0637,
      "step": 2170
    },
    {
      "epoch": 2.75,
      "learning_rate": 3.0370705244122967e-05,
      "loss": 0.0261,
      "step": 2171
    },
    {
      "epoch": 2.75,
      "learning_rate": 3.0361663652802895e-05,
      "loss": 0.3131,
      "step": 2172
    },
    {
      "epoch": 2.75,
      "learning_rate": 3.0352622061482823e-05,
      "loss": 0.0313,
      "step": 2173
    },
    {
      "epoch": 2.75,
      "learning_rate": 3.034358047016275e-05,
      "loss": 0.0108,
      "step": 2174
    },
    {
      "epoch": 2.75,
      "learning_rate": 3.0334538878842676e-05,
      "loss": 0.1148,
      "step": 2175
    },
    {
      "epoch": 2.75,
      "learning_rate": 3.0325497287522605e-05,
      "loss": 0.2248,
      "step": 2176
    },
    {
      "epoch": 2.76,
      "learning_rate": 3.0316455696202533e-05,
      "loss": 0.2357,
      "step": 2177
    },
    {
      "epoch": 2.76,
      "learning_rate": 3.0307414104882458e-05,
      "loss": 0.031,
      "step": 2178
    },
    {
      "epoch": 2.76,
      "learning_rate": 3.029837251356239e-05,
      "loss": 0.0733,
      "step": 2179
    },
    {
      "epoch": 2.76,
      "learning_rate": 3.0289330922242314e-05,
      "loss": 0.0189,
      "step": 2180
    },
    {
      "epoch": 2.76,
      "learning_rate": 3.0280289330922246e-05,
      "loss": 0.0827,
      "step": 2181
    },
    {
      "epoch": 2.76,
      "learning_rate": 3.027124773960217e-05,
      "loss": 0.1288,
      "step": 2182
    },
    {
      "epoch": 2.76,
      "learning_rate": 3.0262206148282102e-05,
      "loss": 0.2135,
      "step": 2183
    },
    {
      "epoch": 2.76,
      "learning_rate": 3.0253164556962027e-05,
      "loss": 0.1624,
      "step": 2184
    },
    {
      "epoch": 2.77,
      "learning_rate": 3.0244122965641952e-05,
      "loss": 0.4355,
      "step": 2185
    },
    {
      "epoch": 2.77,
      "learning_rate": 3.0235081374321884e-05,
      "loss": 0.1548,
      "step": 2186
    },
    {
      "epoch": 2.77,
      "learning_rate": 3.022603978300181e-05,
      "loss": 0.0072,
      "step": 2187
    },
    {
      "epoch": 2.77,
      "learning_rate": 3.021699819168174e-05,
      "loss": 0.2082,
      "step": 2188
    },
    {
      "epoch": 2.77,
      "learning_rate": 3.0207956600361665e-05,
      "loss": 0.2652,
      "step": 2189
    },
    {
      "epoch": 2.77,
      "learning_rate": 3.0198915009041596e-05,
      "loss": 0.3177,
      "step": 2190
    },
    {
      "epoch": 2.77,
      "learning_rate": 3.018987341772152e-05,
      "loss": 0.1585,
      "step": 2191
    },
    {
      "epoch": 2.77,
      "learning_rate": 3.0180831826401446e-05,
      "loss": 0.0487,
      "step": 2192
    },
    {
      "epoch": 2.78,
      "learning_rate": 3.0171790235081378e-05,
      "loss": 0.0994,
      "step": 2193
    },
    {
      "epoch": 2.78,
      "learning_rate": 3.0162748643761303e-05,
      "loss": 0.0596,
      "step": 2194
    },
    {
      "epoch": 2.78,
      "learning_rate": 3.015370705244123e-05,
      "loss": 0.1352,
      "step": 2195
    },
    {
      "epoch": 2.78,
      "learning_rate": 3.014466546112116e-05,
      "loss": 0.0551,
      "step": 2196
    },
    {
      "epoch": 2.78,
      "learning_rate": 3.0135623869801087e-05,
      "loss": 0.1923,
      "step": 2197
    },
    {
      "epoch": 2.78,
      "learning_rate": 3.0126582278481012e-05,
      "loss": 0.0726,
      "step": 2198
    },
    {
      "epoch": 2.78,
      "learning_rate": 3.0117540687160944e-05,
      "loss": 0.2671,
      "step": 2199
    },
    {
      "epoch": 2.78,
      "learning_rate": 3.010849909584087e-05,
      "loss": 0.169,
      "step": 2200
    },
    {
      "epoch": 2.79,
      "learning_rate": 3.0099457504520794e-05,
      "loss": 0.1829,
      "step": 2201
    },
    {
      "epoch": 2.79,
      "learning_rate": 3.0090415913200725e-05,
      "loss": 0.1012,
      "step": 2202
    },
    {
      "epoch": 2.79,
      "learning_rate": 3.008137432188065e-05,
      "loss": 0.471,
      "step": 2203
    },
    {
      "epoch": 2.79,
      "learning_rate": 3.007233273056058e-05,
      "loss": 0.0058,
      "step": 2204
    },
    {
      "epoch": 2.79,
      "learning_rate": 3.0063291139240506e-05,
      "loss": 0.1386,
      "step": 2205
    },
    {
      "epoch": 2.79,
      "learning_rate": 3.0054249547920438e-05,
      "loss": 0.0728,
      "step": 2206
    },
    {
      "epoch": 2.79,
      "learning_rate": 3.0045207956600363e-05,
      "loss": 0.0817,
      "step": 2207
    },
    {
      "epoch": 2.79,
      "learning_rate": 3.0036166365280288e-05,
      "loss": 0.1509,
      "step": 2208
    },
    {
      "epoch": 2.8,
      "learning_rate": 3.002712477396022e-05,
      "loss": 0.1246,
      "step": 2209
    },
    {
      "epoch": 2.8,
      "learning_rate": 3.0018083182640144e-05,
      "loss": 0.1348,
      "step": 2210
    },
    {
      "epoch": 2.8,
      "learning_rate": 3.0009041591320076e-05,
      "loss": 0.1156,
      "step": 2211
    },
    {
      "epoch": 2.8,
      "learning_rate": 3e-05,
      "loss": 0.1034,
      "step": 2212
    },
    {
      "epoch": 2.8,
      "learning_rate": 2.9990958408679932e-05,
      "loss": 0.0652,
      "step": 2213
    },
    {
      "epoch": 2.8,
      "learning_rate": 2.9981916817359857e-05,
      "loss": 0.3507,
      "step": 2214
    },
    {
      "epoch": 2.8,
      "learning_rate": 2.9972875226039785e-05,
      "loss": 0.0816,
      "step": 2215
    },
    {
      "epoch": 2.8,
      "learning_rate": 2.9963833634719714e-05,
      "loss": 0.1148,
      "step": 2216
    },
    {
      "epoch": 2.81,
      "learning_rate": 2.995479204339964e-05,
      "loss": 0.0675,
      "step": 2217
    },
    {
      "epoch": 2.81,
      "learning_rate": 2.9945750452079567e-05,
      "loss": 0.1934,
      "step": 2218
    },
    {
      "epoch": 2.81,
      "learning_rate": 2.9936708860759495e-05,
      "loss": 0.0747,
      "step": 2219
    },
    {
      "epoch": 2.81,
      "learning_rate": 2.9927667269439423e-05,
      "loss": 0.0661,
      "step": 2220
    },
    {
      "epoch": 2.81,
      "learning_rate": 2.9918625678119348e-05,
      "loss": 0.0955,
      "step": 2221
    },
    {
      "epoch": 2.81,
      "learning_rate": 2.990958408679928e-05,
      "loss": 0.1672,
      "step": 2222
    },
    {
      "epoch": 2.81,
      "learning_rate": 2.9900542495479205e-05,
      "loss": 0.0097,
      "step": 2223
    },
    {
      "epoch": 2.81,
      "learning_rate": 2.989150090415913e-05,
      "loss": 0.0813,
      "step": 2224
    },
    {
      "epoch": 2.82,
      "learning_rate": 2.988245931283906e-05,
      "loss": 0.186,
      "step": 2225
    },
    {
      "epoch": 2.82,
      "learning_rate": 2.9873417721518986e-05,
      "loss": 0.0482,
      "step": 2226
    },
    {
      "epoch": 2.82,
      "learning_rate": 2.9864376130198918e-05,
      "loss": 0.138,
      "step": 2227
    },
    {
      "epoch": 2.82,
      "learning_rate": 2.9855334538878842e-05,
      "loss": 0.0697,
      "step": 2228
    },
    {
      "epoch": 2.82,
      "learning_rate": 2.9846292947558774e-05,
      "loss": 0.055,
      "step": 2229
    },
    {
      "epoch": 2.82,
      "learning_rate": 2.98372513562387e-05,
      "loss": 0.0881,
      "step": 2230
    },
    {
      "epoch": 2.82,
      "learning_rate": 2.982820976491863e-05,
      "loss": 0.1717,
      "step": 2231
    },
    {
      "epoch": 2.82,
      "learning_rate": 2.9819168173598555e-05,
      "loss": 0.0559,
      "step": 2232
    },
    {
      "epoch": 2.83,
      "learning_rate": 2.981012658227848e-05,
      "loss": 0.1004,
      "step": 2233
    },
    {
      "epoch": 2.83,
      "learning_rate": 2.9801084990958412e-05,
      "loss": 0.0169,
      "step": 2234
    },
    {
      "epoch": 2.83,
      "learning_rate": 2.9792043399638337e-05,
      "loss": 0.0983,
      "step": 2235
    },
    {
      "epoch": 2.83,
      "learning_rate": 2.9783001808318268e-05,
      "loss": 0.0685,
      "step": 2236
    },
    {
      "epoch": 2.83,
      "learning_rate": 2.9773960216998193e-05,
      "loss": 0.0221,
      "step": 2237
    },
    {
      "epoch": 2.83,
      "learning_rate": 2.9764918625678125e-05,
      "loss": 0.0487,
      "step": 2238
    },
    {
      "epoch": 2.83,
      "learning_rate": 2.975587703435805e-05,
      "loss": 0.2516,
      "step": 2239
    },
    {
      "epoch": 2.83,
      "learning_rate": 2.9746835443037974e-05,
      "loss": 0.0328,
      "step": 2240
    },
    {
      "epoch": 2.84,
      "learning_rate": 2.9737793851717906e-05,
      "loss": 0.2509,
      "step": 2241
    },
    {
      "epoch": 2.84,
      "learning_rate": 2.972875226039783e-05,
      "loss": 0.1051,
      "step": 2242
    },
    {
      "epoch": 2.84,
      "learning_rate": 2.971971066907776e-05,
      "loss": 0.0313,
      "step": 2243
    },
    {
      "epoch": 2.84,
      "learning_rate": 2.9710669077757684e-05,
      "loss": 0.0254,
      "step": 2244
    },
    {
      "epoch": 2.84,
      "learning_rate": 2.9701627486437616e-05,
      "loss": 0.0458,
      "step": 2245
    },
    {
      "epoch": 2.84,
      "learning_rate": 2.969258589511754e-05,
      "loss": 0.0384,
      "step": 2246
    },
    {
      "epoch": 2.84,
      "learning_rate": 2.9683544303797472e-05,
      "loss": 0.1138,
      "step": 2247
    },
    {
      "epoch": 2.85,
      "learning_rate": 2.9674502712477397e-05,
      "loss": 0.184,
      "step": 2248
    },
    {
      "epoch": 2.85,
      "learning_rate": 2.9665461121157322e-05,
      "loss": 0.041,
      "step": 2249
    },
    {
      "epoch": 2.85,
      "learning_rate": 2.9656419529837253e-05,
      "loss": 0.1334,
      "step": 2250
    },
    {
      "epoch": 2.85,
      "eval_accuracy": 0.9069069069069069,
      "eval_loss": 0.2968127131462097,
      "eval_runtime": 45.1676,
      "eval_samples_per_second": 29.49,
      "eval_steps_per_second": 3.697,
      "step": 2250
    },
    {
      "epoch": 2.85,
      "learning_rate": 2.964737793851718e-05,
      "loss": 0.1781,
      "step": 2251
    },
    {
      "epoch": 2.85,
      "learning_rate": 2.963833634719711e-05,
      "loss": 0.1652,
      "step": 2252
    },
    {
      "epoch": 2.85,
      "learning_rate": 2.9629294755877035e-05,
      "loss": 0.0757,
      "step": 2253
    },
    {
      "epoch": 2.85,
      "learning_rate": 2.9620253164556966e-05,
      "loss": 0.0222,
      "step": 2254
    },
    {
      "epoch": 2.85,
      "learning_rate": 2.961121157323689e-05,
      "loss": 0.3486,
      "step": 2255
    },
    {
      "epoch": 2.86,
      "learning_rate": 2.9602169981916816e-05,
      "loss": 0.0751,
      "step": 2256
    },
    {
      "epoch": 2.86,
      "learning_rate": 2.9593128390596748e-05,
      "loss": 0.02,
      "step": 2257
    },
    {
      "epoch": 2.86,
      "learning_rate": 2.9584086799276673e-05,
      "loss": 0.0896,
      "step": 2258
    },
    {
      "epoch": 2.86,
      "learning_rate": 2.9575045207956604e-05,
      "loss": 0.0825,
      "step": 2259
    },
    {
      "epoch": 2.86,
      "learning_rate": 2.956600361663653e-05,
      "loss": 0.1136,
      "step": 2260
    },
    {
      "epoch": 2.86,
      "learning_rate": 2.955696202531646e-05,
      "loss": 0.0514,
      "step": 2261
    },
    {
      "epoch": 2.86,
      "learning_rate": 2.9547920433996386e-05,
      "loss": 0.0323,
      "step": 2262
    },
    {
      "epoch": 2.86,
      "learning_rate": 2.953887884267631e-05,
      "loss": 0.0674,
      "step": 2263
    },
    {
      "epoch": 2.87,
      "learning_rate": 2.9529837251356242e-05,
      "loss": 0.1992,
      "step": 2264
    },
    {
      "epoch": 2.87,
      "learning_rate": 2.9520795660036167e-05,
      "loss": 0.046,
      "step": 2265
    },
    {
      "epoch": 2.87,
      "learning_rate": 2.9511754068716095e-05,
      "loss": 0.014,
      "step": 2266
    },
    {
      "epoch": 2.87,
      "learning_rate": 2.9502712477396023e-05,
      "loss": 0.0752,
      "step": 2267
    },
    {
      "epoch": 2.87,
      "learning_rate": 2.949367088607595e-05,
      "loss": 0.0623,
      "step": 2268
    },
    {
      "epoch": 2.87,
      "learning_rate": 2.9484629294755876e-05,
      "loss": 0.173,
      "step": 2269
    },
    {
      "epoch": 2.87,
      "learning_rate": 2.9475587703435808e-05,
      "loss": 0.0489,
      "step": 2270
    },
    {
      "epoch": 2.87,
      "learning_rate": 2.9466546112115733e-05,
      "loss": 0.1615,
      "step": 2271
    },
    {
      "epoch": 2.88,
      "learning_rate": 2.9457504520795658e-05,
      "loss": 0.3461,
      "step": 2272
    },
    {
      "epoch": 2.88,
      "learning_rate": 2.944846292947559e-05,
      "loss": 0.0788,
      "step": 2273
    },
    {
      "epoch": 2.88,
      "learning_rate": 2.9439421338155514e-05,
      "loss": 0.6983,
      "step": 2274
    },
    {
      "epoch": 2.88,
      "learning_rate": 2.9430379746835446e-05,
      "loss": 0.2654,
      "step": 2275
    },
    {
      "epoch": 2.88,
      "learning_rate": 2.942133815551537e-05,
      "loss": 0.0238,
      "step": 2276
    },
    {
      "epoch": 2.88,
      "learning_rate": 2.9412296564195302e-05,
      "loss": 0.1712,
      "step": 2277
    },
    {
      "epoch": 2.88,
      "learning_rate": 2.9403254972875227e-05,
      "loss": 0.04,
      "step": 2278
    },
    {
      "epoch": 2.88,
      "learning_rate": 2.9394213381555152e-05,
      "loss": 0.0798,
      "step": 2279
    },
    {
      "epoch": 2.89,
      "learning_rate": 2.9385171790235084e-05,
      "loss": 0.012,
      "step": 2280
    },
    {
      "epoch": 2.89,
      "learning_rate": 2.937613019891501e-05,
      "loss": 0.0853,
      "step": 2281
    },
    {
      "epoch": 2.89,
      "learning_rate": 2.936708860759494e-05,
      "loss": 0.121,
      "step": 2282
    },
    {
      "epoch": 2.89,
      "learning_rate": 2.9358047016274865e-05,
      "loss": 0.0958,
      "step": 2283
    },
    {
      "epoch": 2.89,
      "learning_rate": 2.9349005424954797e-05,
      "loss": 0.0166,
      "step": 2284
    },
    {
      "epoch": 2.89,
      "learning_rate": 2.933996383363472e-05,
      "loss": 0.0126,
      "step": 2285
    },
    {
      "epoch": 2.89,
      "learning_rate": 2.933092224231465e-05,
      "loss": 0.1483,
      "step": 2286
    },
    {
      "epoch": 2.89,
      "learning_rate": 2.9321880650994578e-05,
      "loss": 0.2311,
      "step": 2287
    },
    {
      "epoch": 2.9,
      "learning_rate": 2.9312839059674503e-05,
      "loss": 0.0305,
      "step": 2288
    },
    {
      "epoch": 2.9,
      "learning_rate": 2.930379746835443e-05,
      "loss": 0.1662,
      "step": 2289
    },
    {
      "epoch": 2.9,
      "learning_rate": 2.929475587703436e-05,
      "loss": 0.2272,
      "step": 2290
    },
    {
      "epoch": 2.9,
      "learning_rate": 2.9285714285714288e-05,
      "loss": 0.0477,
      "step": 2291
    },
    {
      "epoch": 2.9,
      "learning_rate": 2.9276672694394212e-05,
      "loss": 0.1549,
      "step": 2292
    },
    {
      "epoch": 2.9,
      "learning_rate": 2.9267631103074144e-05,
      "loss": 0.0122,
      "step": 2293
    },
    {
      "epoch": 2.9,
      "learning_rate": 2.925858951175407e-05,
      "loss": 0.2309,
      "step": 2294
    },
    {
      "epoch": 2.9,
      "learning_rate": 2.9249547920433994e-05,
      "loss": 0.1037,
      "step": 2295
    },
    {
      "epoch": 2.91,
      "learning_rate": 2.9240506329113925e-05,
      "loss": 0.2907,
      "step": 2296
    },
    {
      "epoch": 2.91,
      "learning_rate": 2.923146473779385e-05,
      "loss": 0.166,
      "step": 2297
    },
    {
      "epoch": 2.91,
      "learning_rate": 2.9222423146473782e-05,
      "loss": 0.0627,
      "step": 2298
    },
    {
      "epoch": 2.91,
      "learning_rate": 2.9213381555153707e-05,
      "loss": 0.152,
      "step": 2299
    },
    {
      "epoch": 2.91,
      "learning_rate": 2.9204339963833638e-05,
      "loss": 0.0375,
      "step": 2300
    },
    {
      "epoch": 2.91,
      "learning_rate": 2.9195298372513563e-05,
      "loss": 0.0759,
      "step": 2301
    },
    {
      "epoch": 2.91,
      "learning_rate": 2.9186256781193495e-05,
      "loss": 0.0732,
      "step": 2302
    },
    {
      "epoch": 2.91,
      "learning_rate": 2.917721518987342e-05,
      "loss": 0.2542,
      "step": 2303
    },
    {
      "epoch": 2.92,
      "learning_rate": 2.9168173598553344e-05,
      "loss": 0.1337,
      "step": 2304
    },
    {
      "epoch": 2.92,
      "learning_rate": 2.9159132007233276e-05,
      "loss": 0.0406,
      "step": 2305
    },
    {
      "epoch": 2.92,
      "learning_rate": 2.91500904159132e-05,
      "loss": 0.1692,
      "step": 2306
    },
    {
      "epoch": 2.92,
      "learning_rate": 2.9141048824593133e-05,
      "loss": 0.1199,
      "step": 2307
    },
    {
      "epoch": 2.92,
      "learning_rate": 2.9132007233273057e-05,
      "loss": 0.1055,
      "step": 2308
    },
    {
      "epoch": 2.92,
      "learning_rate": 2.9122965641952986e-05,
      "loss": 0.1327,
      "step": 2309
    },
    {
      "epoch": 2.92,
      "learning_rate": 2.9113924050632914e-05,
      "loss": 0.0114,
      "step": 2310
    },
    {
      "epoch": 2.92,
      "learning_rate": 2.910488245931284e-05,
      "loss": 0.166,
      "step": 2311
    },
    {
      "epoch": 2.93,
      "learning_rate": 2.9095840867992767e-05,
      "loss": 0.0545,
      "step": 2312
    },
    {
      "epoch": 2.93,
      "learning_rate": 2.9086799276672695e-05,
      "loss": 0.083,
      "step": 2313
    },
    {
      "epoch": 2.93,
      "learning_rate": 2.9077757685352623e-05,
      "loss": 0.1648,
      "step": 2314
    },
    {
      "epoch": 2.93,
      "learning_rate": 2.9068716094032548e-05,
      "loss": 0.0706,
      "step": 2315
    },
    {
      "epoch": 2.93,
      "learning_rate": 2.905967450271248e-05,
      "loss": 0.2097,
      "step": 2316
    },
    {
      "epoch": 2.93,
      "learning_rate": 2.9050632911392405e-05,
      "loss": 0.0784,
      "step": 2317
    },
    {
      "epoch": 2.93,
      "learning_rate": 2.9041591320072336e-05,
      "loss": 0.0462,
      "step": 2318
    },
    {
      "epoch": 2.93,
      "learning_rate": 2.903254972875226e-05,
      "loss": 0.065,
      "step": 2319
    },
    {
      "epoch": 2.94,
      "learning_rate": 2.9023508137432186e-05,
      "loss": 0.0116,
      "step": 2320
    },
    {
      "epoch": 2.94,
      "learning_rate": 2.9014466546112118e-05,
      "loss": 0.1326,
      "step": 2321
    },
    {
      "epoch": 2.94,
      "learning_rate": 2.9005424954792043e-05,
      "loss": 0.1735,
      "step": 2322
    },
    {
      "epoch": 2.94,
      "learning_rate": 2.8996383363471974e-05,
      "loss": 0.1434,
      "step": 2323
    },
    {
      "epoch": 2.94,
      "learning_rate": 2.89873417721519e-05,
      "loss": 0.0311,
      "step": 2324
    },
    {
      "epoch": 2.94,
      "learning_rate": 2.897830018083183e-05,
      "loss": 0.0649,
      "step": 2325
    },
    {
      "epoch": 2.94,
      "learning_rate": 2.8969258589511756e-05,
      "loss": 0.2623,
      "step": 2326
    },
    {
      "epoch": 2.94,
      "learning_rate": 2.896021699819168e-05,
      "loss": 0.0328,
      "step": 2327
    },
    {
      "epoch": 2.95,
      "learning_rate": 2.8951175406871612e-05,
      "loss": 0.0672,
      "step": 2328
    },
    {
      "epoch": 2.95,
      "learning_rate": 2.8942133815551537e-05,
      "loss": 0.2079,
      "step": 2329
    },
    {
      "epoch": 2.95,
      "learning_rate": 2.893309222423147e-05,
      "loss": 0.1671,
      "step": 2330
    },
    {
      "epoch": 2.95,
      "learning_rate": 2.8924050632911393e-05,
      "loss": 0.1476,
      "step": 2331
    },
    {
      "epoch": 2.95,
      "learning_rate": 2.8915009041591325e-05,
      "loss": 0.2567,
      "step": 2332
    },
    {
      "epoch": 2.95,
      "learning_rate": 2.890596745027125e-05,
      "loss": 0.0553,
      "step": 2333
    },
    {
      "epoch": 2.95,
      "learning_rate": 2.8896925858951178e-05,
      "loss": 0.113,
      "step": 2334
    },
    {
      "epoch": 2.96,
      "learning_rate": 2.8887884267631103e-05,
      "loss": 0.1622,
      "step": 2335
    },
    {
      "epoch": 2.96,
      "learning_rate": 2.887884267631103e-05,
      "loss": 0.0392,
      "step": 2336
    },
    {
      "epoch": 2.96,
      "learning_rate": 2.886980108499096e-05,
      "loss": 0.127,
      "step": 2337
    },
    {
      "epoch": 2.96,
      "learning_rate": 2.8860759493670884e-05,
      "loss": 0.0411,
      "step": 2338
    },
    {
      "epoch": 2.96,
      "learning_rate": 2.8851717902350816e-05,
      "loss": 0.0494,
      "step": 2339
    },
    {
      "epoch": 2.96,
      "learning_rate": 2.884267631103074e-05,
      "loss": 0.0401,
      "step": 2340
    },
    {
      "epoch": 2.96,
      "learning_rate": 2.8833634719710672e-05,
      "loss": 0.1822,
      "step": 2341
    },
    {
      "epoch": 2.96,
      "learning_rate": 2.8824593128390597e-05,
      "loss": 0.0898,
      "step": 2342
    },
    {
      "epoch": 2.97,
      "learning_rate": 2.8815551537070522e-05,
      "loss": 0.0518,
      "step": 2343
    },
    {
      "epoch": 2.97,
      "learning_rate": 2.8806509945750454e-05,
      "loss": 0.137,
      "step": 2344
    },
    {
      "epoch": 2.97,
      "learning_rate": 2.879746835443038e-05,
      "loss": 0.1615,
      "step": 2345
    },
    {
      "epoch": 2.97,
      "learning_rate": 2.878842676311031e-05,
      "loss": 0.0173,
      "step": 2346
    },
    {
      "epoch": 2.97,
      "learning_rate": 2.8779385171790235e-05,
      "loss": 0.3044,
      "step": 2347
    },
    {
      "epoch": 2.97,
      "learning_rate": 2.8770343580470167e-05,
      "loss": 0.1546,
      "step": 2348
    },
    {
      "epoch": 2.97,
      "learning_rate": 2.876130198915009e-05,
      "loss": 0.2143,
      "step": 2349
    },
    {
      "epoch": 2.97,
      "learning_rate": 2.8752260397830023e-05,
      "loss": 0.1105,
      "step": 2350
    },
    {
      "epoch": 2.98,
      "learning_rate": 2.8743218806509948e-05,
      "loss": 0.1448,
      "step": 2351
    },
    {
      "epoch": 2.98,
      "learning_rate": 2.8734177215189873e-05,
      "loss": 0.1281,
      "step": 2352
    },
    {
      "epoch": 2.98,
      "learning_rate": 2.8725135623869804e-05,
      "loss": 0.163,
      "step": 2353
    },
    {
      "epoch": 2.98,
      "learning_rate": 2.871609403254973e-05,
      "loss": 0.0155,
      "step": 2354
    },
    {
      "epoch": 2.98,
      "learning_rate": 2.870705244122966e-05,
      "loss": 0.0074,
      "step": 2355
    },
    {
      "epoch": 2.98,
      "learning_rate": 2.8698010849909586e-05,
      "loss": 0.0921,
      "step": 2356
    },
    {
      "epoch": 2.98,
      "learning_rate": 2.8688969258589514e-05,
      "loss": 0.1951,
      "step": 2357
    },
    {
      "epoch": 2.98,
      "learning_rate": 2.8679927667269442e-05,
      "loss": 0.0856,
      "step": 2358
    },
    {
      "epoch": 2.99,
      "learning_rate": 2.8670886075949367e-05,
      "loss": 0.0818,
      "step": 2359
    },
    {
      "epoch": 2.99,
      "learning_rate": 2.8661844484629295e-05,
      "loss": 0.0802,
      "step": 2360
    },
    {
      "epoch": 2.99,
      "learning_rate": 2.8652802893309223e-05,
      "loss": 0.1466,
      "step": 2361
    },
    {
      "epoch": 2.99,
      "learning_rate": 2.8643761301989152e-05,
      "loss": 0.0293,
      "step": 2362
    },
    {
      "epoch": 2.99,
      "learning_rate": 2.8634719710669077e-05,
      "loss": 0.1567,
      "step": 2363
    },
    {
      "epoch": 2.99,
      "learning_rate": 2.8625678119349008e-05,
      "loss": 0.2453,
      "step": 2364
    },
    {
      "epoch": 2.99,
      "learning_rate": 2.8616636528028933e-05,
      "loss": 0.0698,
      "step": 2365
    },
    {
      "epoch": 2.99,
      "learning_rate": 2.8607594936708865e-05,
      "loss": 0.0765,
      "step": 2366
    },
    {
      "epoch": 3.0,
      "learning_rate": 2.859855334538879e-05,
      "loss": 0.2049,
      "step": 2367
    },
    {
      "epoch": 3.0,
      "learning_rate": 2.8589511754068714e-05,
      "loss": 0.0387,
      "step": 2368
    },
    {
      "epoch": 3.0,
      "learning_rate": 2.8580470162748646e-05,
      "loss": 0.0446,
      "step": 2369
    },
    {
      "epoch": 3.0,
      "learning_rate": 2.857142857142857e-05,
      "loss": 0.2851,
      "step": 2370
    },
    {
      "epoch": 3.0,
      "learning_rate": 2.8562386980108502e-05,
      "loss": 0.2678,
      "step": 2371
    },
    {
      "epoch": 3.0,
      "learning_rate": 2.8553345388788427e-05,
      "loss": 0.0183,
      "step": 2372
    },
    {
      "epoch": 3.0,
      "learning_rate": 2.854430379746836e-05,
      "loss": 0.0205,
      "step": 2373
    },
    {
      "epoch": 3.01,
      "learning_rate": 2.8535262206148284e-05,
      "loss": 0.0508,
      "step": 2374
    },
    {
      "epoch": 3.01,
      "learning_rate": 2.852622061482821e-05,
      "loss": 0.0816,
      "step": 2375
    },
    {
      "epoch": 3.01,
      "learning_rate": 2.851717902350814e-05,
      "loss": 0.007,
      "step": 2376
    },
    {
      "epoch": 3.01,
      "learning_rate": 2.8508137432188065e-05,
      "loss": 0.0603,
      "step": 2377
    },
    {
      "epoch": 3.01,
      "learning_rate": 2.8499095840867997e-05,
      "loss": 0.0346,
      "step": 2378
    },
    {
      "epoch": 3.01,
      "learning_rate": 2.849005424954792e-05,
      "loss": 0.0247,
      "step": 2379
    },
    {
      "epoch": 3.01,
      "learning_rate": 2.848101265822785e-05,
      "loss": 0.0146,
      "step": 2380
    },
    {
      "epoch": 3.01,
      "learning_rate": 2.8471971066907778e-05,
      "loss": 0.0165,
      "step": 2381
    },
    {
      "epoch": 3.02,
      "learning_rate": 2.8462929475587706e-05,
      "loss": 0.0162,
      "step": 2382
    },
    {
      "epoch": 3.02,
      "learning_rate": 2.845388788426763e-05,
      "loss": 0.0082,
      "step": 2383
    },
    {
      "epoch": 3.02,
      "learning_rate": 2.844484629294756e-05,
      "loss": 0.0113,
      "step": 2384
    },
    {
      "epoch": 3.02,
      "learning_rate": 2.8435804701627488e-05,
      "loss": 0.049,
      "step": 2385
    },
    {
      "epoch": 3.02,
      "learning_rate": 2.8426763110307413e-05,
      "loss": 0.0324,
      "step": 2386
    },
    {
      "epoch": 3.02,
      "learning_rate": 2.8417721518987344e-05,
      "loss": 0.0726,
      "step": 2387
    },
    {
      "epoch": 3.02,
      "learning_rate": 2.840867992766727e-05,
      "loss": 0.006,
      "step": 2388
    },
    {
      "epoch": 3.02,
      "learning_rate": 2.83996383363472e-05,
      "loss": 0.0162,
      "step": 2389
    },
    {
      "epoch": 3.03,
      "learning_rate": 2.8390596745027125e-05,
      "loss": 0.0199,
      "step": 2390
    },
    {
      "epoch": 3.03,
      "learning_rate": 2.838155515370705e-05,
      "loss": 0.0117,
      "step": 2391
    },
    {
      "epoch": 3.03,
      "learning_rate": 2.8372513562386982e-05,
      "loss": 0.0061,
      "step": 2392
    },
    {
      "epoch": 3.03,
      "learning_rate": 2.8363471971066907e-05,
      "loss": 0.1036,
      "step": 2393
    },
    {
      "epoch": 3.03,
      "learning_rate": 2.835443037974684e-05,
      "loss": 0.1641,
      "step": 2394
    },
    {
      "epoch": 3.03,
      "learning_rate": 2.8345388788426763e-05,
      "loss": 0.0212,
      "step": 2395
    },
    {
      "epoch": 3.03,
      "learning_rate": 2.8336347197106695e-05,
      "loss": 0.0244,
      "step": 2396
    },
    {
      "epoch": 3.03,
      "learning_rate": 2.832730560578662e-05,
      "loss": 0.0272,
      "step": 2397
    },
    {
      "epoch": 3.04,
      "learning_rate": 2.831826401446655e-05,
      "loss": 0.157,
      "step": 2398
    },
    {
      "epoch": 3.04,
      "learning_rate": 2.8309222423146476e-05,
      "loss": 0.0964,
      "step": 2399
    },
    {
      "epoch": 3.04,
      "learning_rate": 2.83001808318264e-05,
      "loss": 0.0202,
      "step": 2400
    },
    {
      "epoch": 3.04,
      "learning_rate": 2.8291139240506333e-05,
      "loss": 0.0496,
      "step": 2401
    },
    {
      "epoch": 3.04,
      "learning_rate": 2.8282097649186258e-05,
      "loss": 0.0064,
      "step": 2402
    },
    {
      "epoch": 3.04,
      "learning_rate": 2.8273056057866186e-05,
      "loss": 0.0665,
      "step": 2403
    },
    {
      "epoch": 3.04,
      "learning_rate": 2.8264014466546114e-05,
      "loss": 0.156,
      "step": 2404
    },
    {
      "epoch": 3.04,
      "learning_rate": 2.8254972875226042e-05,
      "loss": 0.027,
      "step": 2405
    },
    {
      "epoch": 3.05,
      "learning_rate": 2.8245931283905967e-05,
      "loss": 0.0036,
      "step": 2406
    },
    {
      "epoch": 3.05,
      "learning_rate": 2.8236889692585895e-05,
      "loss": 0.034,
      "step": 2407
    },
    {
      "epoch": 3.05,
      "learning_rate": 2.8227848101265824e-05,
      "loss": 0.0235,
      "step": 2408
    },
    {
      "epoch": 3.05,
      "learning_rate": 2.821880650994575e-05,
      "loss": 0.0044,
      "step": 2409
    },
    {
      "epoch": 3.05,
      "learning_rate": 2.820976491862568e-05,
      "loss": 0.009,
      "step": 2410
    },
    {
      "epoch": 3.05,
      "learning_rate": 2.8200723327305605e-05,
      "loss": 0.0056,
      "step": 2411
    },
    {
      "epoch": 3.05,
      "learning_rate": 2.8191681735985537e-05,
      "loss": 0.0091,
      "step": 2412
    },
    {
      "epoch": 3.05,
      "learning_rate": 2.818264014466546e-05,
      "loss": 0.0099,
      "step": 2413
    },
    {
      "epoch": 3.06,
      "learning_rate": 2.8173598553345393e-05,
      "loss": 0.0954,
      "step": 2414
    },
    {
      "epoch": 3.06,
      "learning_rate": 2.8164556962025318e-05,
      "loss": 0.0064,
      "step": 2415
    },
    {
      "epoch": 3.06,
      "learning_rate": 2.8155515370705243e-05,
      "loss": 0.0047,
      "step": 2416
    },
    {
      "epoch": 3.06,
      "learning_rate": 2.8146473779385174e-05,
      "loss": 0.0641,
      "step": 2417
    },
    {
      "epoch": 3.06,
      "learning_rate": 2.81374321880651e-05,
      "loss": 0.211,
      "step": 2418
    },
    {
      "epoch": 3.06,
      "learning_rate": 2.812839059674503e-05,
      "loss": 0.1375,
      "step": 2419
    },
    {
      "epoch": 3.06,
      "learning_rate": 2.8119349005424956e-05,
      "loss": 0.0551,
      "step": 2420
    },
    {
      "epoch": 3.06,
      "learning_rate": 2.8110307414104887e-05,
      "loss": 0.0541,
      "step": 2421
    },
    {
      "epoch": 3.07,
      "learning_rate": 2.8101265822784812e-05,
      "loss": 0.0741,
      "step": 2422
    },
    {
      "epoch": 3.07,
      "learning_rate": 2.8092224231464737e-05,
      "loss": 0.024,
      "step": 2423
    },
    {
      "epoch": 3.07,
      "learning_rate": 2.808318264014467e-05,
      "loss": 0.0046,
      "step": 2424
    },
    {
      "epoch": 3.07,
      "learning_rate": 2.8074141048824593e-05,
      "loss": 0.0374,
      "step": 2425
    },
    {
      "epoch": 3.07,
      "learning_rate": 2.8065099457504525e-05,
      "loss": 0.0417,
      "step": 2426
    },
    {
      "epoch": 3.07,
      "learning_rate": 2.805605786618445e-05,
      "loss": 0.1237,
      "step": 2427
    },
    {
      "epoch": 3.07,
      "learning_rate": 2.8047016274864378e-05,
      "loss": 0.0304,
      "step": 2428
    },
    {
      "epoch": 3.07,
      "learning_rate": 2.8037974683544303e-05,
      "loss": 0.0311,
      "step": 2429
    },
    {
      "epoch": 3.08,
      "learning_rate": 2.8028933092224235e-05,
      "loss": 0.0029,
      "step": 2430
    },
    {
      "epoch": 3.08,
      "learning_rate": 2.801989150090416e-05,
      "loss": 0.032,
      "step": 2431
    },
    {
      "epoch": 3.08,
      "learning_rate": 2.8010849909584084e-05,
      "loss": 0.0805,
      "step": 2432
    },
    {
      "epoch": 3.08,
      "learning_rate": 2.8001808318264016e-05,
      "loss": 0.0072,
      "step": 2433
    },
    {
      "epoch": 3.08,
      "learning_rate": 2.799276672694394e-05,
      "loss": 0.0099,
      "step": 2434
    },
    {
      "epoch": 3.08,
      "learning_rate": 2.7983725135623872e-05,
      "loss": 0.0118,
      "step": 2435
    },
    {
      "epoch": 3.08,
      "learning_rate": 2.7974683544303797e-05,
      "loss": 0.0048,
      "step": 2436
    },
    {
      "epoch": 3.08,
      "learning_rate": 2.796564195298373e-05,
      "loss": 0.0182,
      "step": 2437
    },
    {
      "epoch": 3.09,
      "learning_rate": 2.7956600361663654e-05,
      "loss": 0.0351,
      "step": 2438
    },
    {
      "epoch": 3.09,
      "learning_rate": 2.794755877034358e-05,
      "loss": 0.016,
      "step": 2439
    },
    {
      "epoch": 3.09,
      "learning_rate": 2.793851717902351e-05,
      "loss": 0.101,
      "step": 2440
    },
    {
      "epoch": 3.09,
      "learning_rate": 2.7929475587703435e-05,
      "loss": 0.0508,
      "step": 2441
    },
    {
      "epoch": 3.09,
      "learning_rate": 2.7920433996383367e-05,
      "loss": 0.0105,
      "step": 2442
    },
    {
      "epoch": 3.09,
      "learning_rate": 2.791139240506329e-05,
      "loss": 0.0088,
      "step": 2443
    },
    {
      "epoch": 3.09,
      "learning_rate": 2.7902350813743223e-05,
      "loss": 0.1138,
      "step": 2444
    },
    {
      "epoch": 3.09,
      "learning_rate": 2.7893309222423148e-05,
      "loss": 0.0842,
      "step": 2445
    },
    {
      "epoch": 3.1,
      "learning_rate": 2.788426763110308e-05,
      "loss": 0.1445,
      "step": 2446
    },
    {
      "epoch": 3.1,
      "learning_rate": 2.7875226039783005e-05,
      "loss": 0.0061,
      "step": 2447
    },
    {
      "epoch": 3.1,
      "learning_rate": 2.786618444846293e-05,
      "loss": 0.1359,
      "step": 2448
    },
    {
      "epoch": 3.1,
      "learning_rate": 2.785714285714286e-05,
      "loss": 0.0239,
      "step": 2449
    },
    {
      "epoch": 3.1,
      "learning_rate": 2.7848101265822786e-05,
      "loss": 0.0079,
      "step": 2450
    },
    {
      "epoch": 3.1,
      "learning_rate": 2.7839059674502714e-05,
      "loss": 0.0362,
      "step": 2451
    },
    {
      "epoch": 3.1,
      "learning_rate": 2.7830018083182642e-05,
      "loss": 0.0418,
      "step": 2452
    },
    {
      "epoch": 3.1,
      "learning_rate": 2.782097649186257e-05,
      "loss": 0.1207,
      "step": 2453
    },
    {
      "epoch": 3.11,
      "learning_rate": 2.7811934900542495e-05,
      "loss": 0.0024,
      "step": 2454
    },
    {
      "epoch": 3.11,
      "learning_rate": 2.7802893309222424e-05,
      "loss": 0.0321,
      "step": 2455
    },
    {
      "epoch": 3.11,
      "learning_rate": 2.7793851717902352e-05,
      "loss": 0.1645,
      "step": 2456
    },
    {
      "epoch": 3.11,
      "learning_rate": 2.7784810126582277e-05,
      "loss": 0.0955,
      "step": 2457
    },
    {
      "epoch": 3.11,
      "learning_rate": 2.777576853526221e-05,
      "loss": 0.0244,
      "step": 2458
    },
    {
      "epoch": 3.11,
      "learning_rate": 2.7766726943942133e-05,
      "loss": 0.004,
      "step": 2459
    },
    {
      "epoch": 3.11,
      "learning_rate": 2.7757685352622065e-05,
      "loss": 0.0084,
      "step": 2460
    },
    {
      "epoch": 3.12,
      "learning_rate": 2.774864376130199e-05,
      "loss": 0.0071,
      "step": 2461
    },
    {
      "epoch": 3.12,
      "learning_rate": 2.773960216998192e-05,
      "loss": 0.0259,
      "step": 2462
    },
    {
      "epoch": 3.12,
      "learning_rate": 2.7730560578661846e-05,
      "loss": 0.0196,
      "step": 2463
    },
    {
      "epoch": 3.12,
      "learning_rate": 2.772151898734177e-05,
      "loss": 0.0965,
      "step": 2464
    },
    {
      "epoch": 3.12,
      "learning_rate": 2.7712477396021703e-05,
      "loss": 0.1466,
      "step": 2465
    },
    {
      "epoch": 3.12,
      "learning_rate": 2.7703435804701627e-05,
      "loss": 0.0265,
      "step": 2466
    },
    {
      "epoch": 3.12,
      "learning_rate": 2.769439421338156e-05,
      "loss": 0.0043,
      "step": 2467
    },
    {
      "epoch": 3.12,
      "learning_rate": 2.7685352622061484e-05,
      "loss": 0.1089,
      "step": 2468
    },
    {
      "epoch": 3.13,
      "learning_rate": 2.7676311030741416e-05,
      "loss": 0.013,
      "step": 2469
    },
    {
      "epoch": 3.13,
      "learning_rate": 2.766726943942134e-05,
      "loss": 0.0063,
      "step": 2470
    },
    {
      "epoch": 3.13,
      "learning_rate": 2.7658227848101265e-05,
      "loss": 0.1297,
      "step": 2471
    },
    {
      "epoch": 3.13,
      "learning_rate": 2.7649186256781197e-05,
      "loss": 0.0182,
      "step": 2472
    },
    {
      "epoch": 3.13,
      "learning_rate": 2.7640144665461122e-05,
      "loss": 0.0087,
      "step": 2473
    },
    {
      "epoch": 3.13,
      "learning_rate": 2.763110307414105e-05,
      "loss": 0.0037,
      "step": 2474
    },
    {
      "epoch": 3.13,
      "learning_rate": 2.7622061482820978e-05,
      "loss": 0.0364,
      "step": 2475
    },
    {
      "epoch": 3.13,
      "learning_rate": 2.7613019891500906e-05,
      "loss": 0.0116,
      "step": 2476
    },
    {
      "epoch": 3.14,
      "learning_rate": 2.760397830018083e-05,
      "loss": 0.0196,
      "step": 2477
    },
    {
      "epoch": 3.14,
      "learning_rate": 2.7594936708860763e-05,
      "loss": 0.0034,
      "step": 2478
    },
    {
      "epoch": 3.14,
      "learning_rate": 2.7585895117540688e-05,
      "loss": 0.0044,
      "step": 2479
    },
    {
      "epoch": 3.14,
      "learning_rate": 2.7576853526220613e-05,
      "loss": 0.0314,
      "step": 2480
    },
    {
      "epoch": 3.14,
      "learning_rate": 2.7567811934900544e-05,
      "loss": 0.0133,
      "step": 2481
    },
    {
      "epoch": 3.14,
      "learning_rate": 2.755877034358047e-05,
      "loss": 0.0094,
      "step": 2482
    },
    {
      "epoch": 3.14,
      "learning_rate": 2.75497287522604e-05,
      "loss": 0.2978,
      "step": 2483
    },
    {
      "epoch": 3.14,
      "learning_rate": 2.7540687160940326e-05,
      "loss": 0.0083,
      "step": 2484
    },
    {
      "epoch": 3.15,
      "learning_rate": 2.7531645569620257e-05,
      "loss": 0.0041,
      "step": 2485
    },
    {
      "epoch": 3.15,
      "learning_rate": 2.7522603978300182e-05,
      "loss": 0.0125,
      "step": 2486
    },
    {
      "epoch": 3.15,
      "learning_rate": 2.7513562386980107e-05,
      "loss": 0.1961,
      "step": 2487
    },
    {
      "epoch": 3.15,
      "learning_rate": 2.750452079566004e-05,
      "loss": 0.0141,
      "step": 2488
    },
    {
      "epoch": 3.15,
      "learning_rate": 2.7495479204339963e-05,
      "loss": 0.1455,
      "step": 2489
    },
    {
      "epoch": 3.15,
      "learning_rate": 2.7486437613019895e-05,
      "loss": 0.0095,
      "step": 2490
    },
    {
      "epoch": 3.15,
      "learning_rate": 2.747739602169982e-05,
      "loss": 0.0075,
      "step": 2491
    },
    {
      "epoch": 3.15,
      "learning_rate": 2.746835443037975e-05,
      "loss": 0.0012,
      "step": 2492
    },
    {
      "epoch": 3.16,
      "learning_rate": 2.7459312839059676e-05,
      "loss": 0.0027,
      "step": 2493
    },
    {
      "epoch": 3.16,
      "learning_rate": 2.7450271247739605e-05,
      "loss": 0.0018,
      "step": 2494
    },
    {
      "epoch": 3.16,
      "learning_rate": 2.7441229656419533e-05,
      "loss": 0.0048,
      "step": 2495
    },
    {
      "epoch": 3.16,
      "learning_rate": 2.7432188065099458e-05,
      "loss": 0.0241,
      "step": 2496
    },
    {
      "epoch": 3.16,
      "learning_rate": 2.7423146473779386e-05,
      "loss": 0.0022,
      "step": 2497
    },
    {
      "epoch": 3.16,
      "learning_rate": 2.7414104882459314e-05,
      "loss": 0.3222,
      "step": 2498
    },
    {
      "epoch": 3.16,
      "learning_rate": 2.7405063291139242e-05,
      "loss": 0.0104,
      "step": 2499
    },
    {
      "epoch": 3.16,
      "learning_rate": 2.7396021699819167e-05,
      "loss": 0.0043,
      "step": 2500
    },
    {
      "epoch": 3.16,
      "eval_accuracy": 0.8933933933933934,
      "eval_loss": 0.3841577172279358,
      "eval_runtime": 45.1205,
      "eval_samples_per_second": 29.521,
      "eval_steps_per_second": 3.701,
      "step": 2500
    },
    {
      "epoch": 3.17,
      "learning_rate": 2.73869801084991e-05,
      "loss": 0.0076,
      "step": 2501
    },
    {
      "epoch": 3.17,
      "learning_rate": 2.7377938517179024e-05,
      "loss": 0.005,
      "step": 2502
    },
    {
      "epoch": 3.17,
      "learning_rate": 2.736889692585895e-05,
      "loss": 0.0041,
      "step": 2503
    },
    {
      "epoch": 3.17,
      "learning_rate": 2.735985533453888e-05,
      "loss": 0.0082,
      "step": 2504
    },
    {
      "epoch": 3.17,
      "learning_rate": 2.7350813743218805e-05,
      "loss": 0.0351,
      "step": 2505
    },
    {
      "epoch": 3.17,
      "learning_rate": 2.7341772151898737e-05,
      "loss": 0.0011,
      "step": 2506
    },
    {
      "epoch": 3.17,
      "learning_rate": 2.733273056057866e-05,
      "loss": 0.0083,
      "step": 2507
    },
    {
      "epoch": 3.17,
      "learning_rate": 2.7323688969258593e-05,
      "loss": 0.0046,
      "step": 2508
    },
    {
      "epoch": 3.18,
      "learning_rate": 2.7314647377938518e-05,
      "loss": 0.0023,
      "step": 2509
    },
    {
      "epoch": 3.18,
      "learning_rate": 2.730560578661845e-05,
      "loss": 0.149,
      "step": 2510
    },
    {
      "epoch": 3.18,
      "learning_rate": 2.7296564195298374e-05,
      "loss": 0.0722,
      "step": 2511
    },
    {
      "epoch": 3.18,
      "learning_rate": 2.72875226039783e-05,
      "loss": 0.0022,
      "step": 2512
    },
    {
      "epoch": 3.18,
      "learning_rate": 2.727848101265823e-05,
      "loss": 0.0056,
      "step": 2513
    },
    {
      "epoch": 3.18,
      "learning_rate": 2.7269439421338156e-05,
      "loss": 0.1072,
      "step": 2514
    },
    {
      "epoch": 3.18,
      "learning_rate": 2.7260397830018087e-05,
      "loss": 0.2314,
      "step": 2515
    },
    {
      "epoch": 3.18,
      "learning_rate": 2.7251356238698012e-05,
      "loss": 0.0203,
      "step": 2516
    },
    {
      "epoch": 3.19,
      "learning_rate": 2.7242314647377944e-05,
      "loss": 0.0043,
      "step": 2517
    },
    {
      "epoch": 3.19,
      "learning_rate": 2.723327305605787e-05,
      "loss": 0.0301,
      "step": 2518
    },
    {
      "epoch": 3.19,
      "learning_rate": 2.7224231464737794e-05,
      "loss": 0.0626,
      "step": 2519
    },
    {
      "epoch": 3.19,
      "learning_rate": 2.7215189873417722e-05,
      "loss": 0.0022,
      "step": 2520
    },
    {
      "epoch": 3.19,
      "learning_rate": 2.720614828209765e-05,
      "loss": 0.1668,
      "step": 2521
    },
    {
      "epoch": 3.19,
      "learning_rate": 2.719710669077758e-05,
      "loss": 0.1115,
      "step": 2522
    },
    {
      "epoch": 3.19,
      "learning_rate": 2.7188065099457503e-05,
      "loss": 0.0995,
      "step": 2523
    },
    {
      "epoch": 3.19,
      "learning_rate": 2.7179023508137435e-05,
      "loss": 0.0152,
      "step": 2524
    },
    {
      "epoch": 3.2,
      "learning_rate": 2.716998191681736e-05,
      "loss": 0.087,
      "step": 2525
    },
    {
      "epoch": 3.2,
      "learning_rate": 2.7160940325497284e-05,
      "loss": 0.0356,
      "step": 2526
    },
    {
      "epoch": 3.2,
      "learning_rate": 2.7151898734177216e-05,
      "loss": 0.1561,
      "step": 2527
    },
    {
      "epoch": 3.2,
      "learning_rate": 2.714285714285714e-05,
      "loss": 0.0157,
      "step": 2528
    },
    {
      "epoch": 3.2,
      "learning_rate": 2.7133815551537073e-05,
      "loss": 0.0144,
      "step": 2529
    },
    {
      "epoch": 3.2,
      "learning_rate": 2.7124773960216997e-05,
      "loss": 0.0157,
      "step": 2530
    },
    {
      "epoch": 3.2,
      "learning_rate": 2.711573236889693e-05,
      "loss": 0.0138,
      "step": 2531
    },
    {
      "epoch": 3.2,
      "learning_rate": 2.7106690777576854e-05,
      "loss": 0.03,
      "step": 2532
    },
    {
      "epoch": 3.21,
      "learning_rate": 2.7097649186256786e-05,
      "loss": 0.0017,
      "step": 2533
    },
    {
      "epoch": 3.21,
      "learning_rate": 2.708860759493671e-05,
      "loss": 0.0901,
      "step": 2534
    },
    {
      "epoch": 3.21,
      "learning_rate": 2.7079566003616635e-05,
      "loss": 0.0155,
      "step": 2535
    },
    {
      "epoch": 3.21,
      "learning_rate": 2.7070524412296567e-05,
      "loss": 0.0035,
      "step": 2536
    },
    {
      "epoch": 3.21,
      "learning_rate": 2.7061482820976492e-05,
      "loss": 0.006,
      "step": 2537
    },
    {
      "epoch": 3.21,
      "learning_rate": 2.7052441229656423e-05,
      "loss": 0.0082,
      "step": 2538
    },
    {
      "epoch": 3.21,
      "learning_rate": 2.7043399638336348e-05,
      "loss": 0.0063,
      "step": 2539
    },
    {
      "epoch": 3.22,
      "learning_rate": 2.703435804701628e-05,
      "loss": 0.0075,
      "step": 2540
    },
    {
      "epoch": 3.22,
      "learning_rate": 2.7025316455696205e-05,
      "loss": 0.017,
      "step": 2541
    },
    {
      "epoch": 3.22,
      "learning_rate": 2.701627486437613e-05,
      "loss": 0.0147,
      "step": 2542
    },
    {
      "epoch": 3.22,
      "learning_rate": 2.700723327305606e-05,
      "loss": 0.1125,
      "step": 2543
    },
    {
      "epoch": 3.22,
      "learning_rate": 2.6998191681735986e-05,
      "loss": 0.008,
      "step": 2544
    },
    {
      "epoch": 3.22,
      "learning_rate": 2.6989150090415914e-05,
      "loss": 0.0508,
      "step": 2545
    },
    {
      "epoch": 3.22,
      "learning_rate": 2.6980108499095842e-05,
      "loss": 0.2573,
      "step": 2546
    },
    {
      "epoch": 3.22,
      "learning_rate": 2.697106690777577e-05,
      "loss": 0.1974,
      "step": 2547
    },
    {
      "epoch": 3.23,
      "learning_rate": 2.6962025316455696e-05,
      "loss": 0.0036,
      "step": 2548
    },
    {
      "epoch": 3.23,
      "learning_rate": 2.6952983725135627e-05,
      "loss": 0.0184,
      "step": 2549
    },
    {
      "epoch": 3.23,
      "learning_rate": 2.6943942133815552e-05,
      "loss": 0.0052,
      "step": 2550
    },
    {
      "epoch": 3.23,
      "learning_rate": 2.6934900542495477e-05,
      "loss": 0.0845,
      "step": 2551
    },
    {
      "epoch": 3.23,
      "learning_rate": 2.692585895117541e-05,
      "loss": 0.009,
      "step": 2552
    },
    {
      "epoch": 3.23,
      "learning_rate": 2.6916817359855333e-05,
      "loss": 0.0037,
      "step": 2553
    },
    {
      "epoch": 3.23,
      "learning_rate": 2.6907775768535265e-05,
      "loss": 0.1169,
      "step": 2554
    },
    {
      "epoch": 3.23,
      "learning_rate": 2.689873417721519e-05,
      "loss": 0.0088,
      "step": 2555
    },
    {
      "epoch": 3.24,
      "learning_rate": 2.688969258589512e-05,
      "loss": 0.0011,
      "step": 2556
    },
    {
      "epoch": 3.24,
      "learning_rate": 2.6880650994575046e-05,
      "loss": 0.0146,
      "step": 2557
    },
    {
      "epoch": 3.24,
      "learning_rate": 2.687160940325497e-05,
      "loss": 0.1325,
      "step": 2558
    },
    {
      "epoch": 3.24,
      "learning_rate": 2.6862567811934903e-05,
      "loss": 0.0272,
      "step": 2559
    },
    {
      "epoch": 3.24,
      "learning_rate": 2.6853526220614828e-05,
      "loss": 0.0073,
      "step": 2560
    },
    {
      "epoch": 3.24,
      "learning_rate": 2.684448462929476e-05,
      "loss": 0.1024,
      "step": 2561
    },
    {
      "epoch": 3.24,
      "learning_rate": 2.6835443037974684e-05,
      "loss": 0.0246,
      "step": 2562
    },
    {
      "epoch": 3.24,
      "learning_rate": 2.6826401446654616e-05,
      "loss": 0.007,
      "step": 2563
    },
    {
      "epoch": 3.25,
      "learning_rate": 2.681735985533454e-05,
      "loss": 0.0803,
      "step": 2564
    },
    {
      "epoch": 3.25,
      "learning_rate": 2.680831826401447e-05,
      "loss": 0.0103,
      "step": 2565
    },
    {
      "epoch": 3.25,
      "learning_rate": 2.6799276672694397e-05,
      "loss": 0.0193,
      "step": 2566
    },
    {
      "epoch": 3.25,
      "learning_rate": 2.6790235081374322e-05,
      "loss": 0.0045,
      "step": 2567
    },
    {
      "epoch": 3.25,
      "learning_rate": 2.678119349005425e-05,
      "loss": 0.2248,
      "step": 2568
    },
    {
      "epoch": 3.25,
      "learning_rate": 2.677215189873418e-05,
      "loss": 0.0107,
      "step": 2569
    },
    {
      "epoch": 3.25,
      "learning_rate": 2.6763110307414107e-05,
      "loss": 0.0028,
      "step": 2570
    },
    {
      "epoch": 3.25,
      "learning_rate": 2.675406871609403e-05,
      "loss": 0.0893,
      "step": 2571
    },
    {
      "epoch": 3.26,
      "learning_rate": 2.6745027124773963e-05,
      "loss": 0.0042,
      "step": 2572
    },
    {
      "epoch": 3.26,
      "learning_rate": 2.6735985533453888e-05,
      "loss": 0.0567,
      "step": 2573
    },
    {
      "epoch": 3.26,
      "learning_rate": 2.6726943942133813e-05,
      "loss": 0.0131,
      "step": 2574
    },
    {
      "epoch": 3.26,
      "learning_rate": 2.6717902350813744e-05,
      "loss": 0.004,
      "step": 2575
    },
    {
      "epoch": 3.26,
      "learning_rate": 2.670886075949367e-05,
      "loss": 0.0036,
      "step": 2576
    },
    {
      "epoch": 3.26,
      "learning_rate": 2.66998191681736e-05,
      "loss": 0.0035,
      "step": 2577
    },
    {
      "epoch": 3.26,
      "learning_rate": 2.6690777576853526e-05,
      "loss": 0.0034,
      "step": 2578
    },
    {
      "epoch": 3.26,
      "learning_rate": 2.6681735985533457e-05,
      "loss": 0.0772,
      "step": 2579
    },
    {
      "epoch": 3.27,
      "learning_rate": 2.6672694394213382e-05,
      "loss": 0.0034,
      "step": 2580
    },
    {
      "epoch": 3.27,
      "learning_rate": 2.6663652802893314e-05,
      "loss": 0.0016,
      "step": 2581
    },
    {
      "epoch": 3.27,
      "learning_rate": 2.665461121157324e-05,
      "loss": 0.0087,
      "step": 2582
    },
    {
      "epoch": 3.27,
      "learning_rate": 2.6645569620253164e-05,
      "loss": 0.0026,
      "step": 2583
    },
    {
      "epoch": 3.27,
      "learning_rate": 2.6636528028933095e-05,
      "loss": 0.0019,
      "step": 2584
    },
    {
      "epoch": 3.27,
      "learning_rate": 2.662748643761302e-05,
      "loss": 0.0017,
      "step": 2585
    },
    {
      "epoch": 3.27,
      "learning_rate": 2.661844484629295e-05,
      "loss": 0.002,
      "step": 2586
    },
    {
      "epoch": 3.27,
      "learning_rate": 2.6609403254972877e-05,
      "loss": 0.0068,
      "step": 2587
    },
    {
      "epoch": 3.28,
      "learning_rate": 2.6600361663652805e-05,
      "loss": 0.0609,
      "step": 2588
    },
    {
      "epoch": 3.28,
      "learning_rate": 2.6591320072332733e-05,
      "loss": 0.0061,
      "step": 2589
    },
    {
      "epoch": 3.28,
      "learning_rate": 2.6582278481012658e-05,
      "loss": 0.004,
      "step": 2590
    },
    {
      "epoch": 3.28,
      "learning_rate": 2.6573236889692586e-05,
      "loss": 0.0032,
      "step": 2591
    },
    {
      "epoch": 3.28,
      "learning_rate": 2.6564195298372514e-05,
      "loss": 0.0016,
      "step": 2592
    },
    {
      "epoch": 3.28,
      "learning_rate": 2.6555153707052443e-05,
      "loss": 0.0426,
      "step": 2593
    },
    {
      "epoch": 3.28,
      "learning_rate": 2.6546112115732367e-05,
      "loss": 0.0026,
      "step": 2594
    },
    {
      "epoch": 3.28,
      "learning_rate": 2.65370705244123e-05,
      "loss": 0.0015,
      "step": 2595
    },
    {
      "epoch": 3.29,
      "learning_rate": 2.6528028933092224e-05,
      "loss": 0.0022,
      "step": 2596
    },
    {
      "epoch": 3.29,
      "learning_rate": 2.6518987341772155e-05,
      "loss": 0.1786,
      "step": 2597
    },
    {
      "epoch": 3.29,
      "learning_rate": 2.650994575045208e-05,
      "loss": 0.1367,
      "step": 2598
    },
    {
      "epoch": 3.29,
      "learning_rate": 2.6500904159132005e-05,
      "loss": 0.1353,
      "step": 2599
    },
    {
      "epoch": 3.29,
      "learning_rate": 2.6491862567811937e-05,
      "loss": 0.0032,
      "step": 2600
    },
    {
      "epoch": 3.29,
      "learning_rate": 2.648282097649186e-05,
      "loss": 0.0443,
      "step": 2601
    },
    {
      "epoch": 3.29,
      "learning_rate": 2.6473779385171793e-05,
      "loss": 0.0083,
      "step": 2602
    },
    {
      "epoch": 3.29,
      "learning_rate": 2.6464737793851718e-05,
      "loss": 0.0122,
      "step": 2603
    },
    {
      "epoch": 3.3,
      "learning_rate": 2.645569620253165e-05,
      "loss": 0.0063,
      "step": 2604
    },
    {
      "epoch": 3.3,
      "learning_rate": 2.6446654611211575e-05,
      "loss": 0.0058,
      "step": 2605
    },
    {
      "epoch": 3.3,
      "learning_rate": 2.64376130198915e-05,
      "loss": 0.0208,
      "step": 2606
    },
    {
      "epoch": 3.3,
      "learning_rate": 2.642857142857143e-05,
      "loss": 0.187,
      "step": 2607
    },
    {
      "epoch": 3.3,
      "learning_rate": 2.6419529837251356e-05,
      "loss": 0.1966,
      "step": 2608
    },
    {
      "epoch": 3.3,
      "learning_rate": 2.6410488245931288e-05,
      "loss": 0.0725,
      "step": 2609
    },
    {
      "epoch": 3.3,
      "learning_rate": 2.6401446654611212e-05,
      "loss": 0.0359,
      "step": 2610
    },
    {
      "epoch": 3.3,
      "learning_rate": 2.6392405063291144e-05,
      "loss": 0.0018,
      "step": 2611
    },
    {
      "epoch": 3.31,
      "learning_rate": 2.638336347197107e-05,
      "loss": 0.012,
      "step": 2612
    },
    {
      "epoch": 3.31,
      "learning_rate": 2.6374321880650997e-05,
      "loss": 0.0097,
      "step": 2613
    },
    {
      "epoch": 3.31,
      "learning_rate": 2.6365280289330922e-05,
      "loss": 0.0784,
      "step": 2614
    },
    {
      "epoch": 3.31,
      "learning_rate": 2.635623869801085e-05,
      "loss": 0.0742,
      "step": 2615
    },
    {
      "epoch": 3.31,
      "learning_rate": 2.634719710669078e-05,
      "loss": 0.0076,
      "step": 2616
    },
    {
      "epoch": 3.31,
      "learning_rate": 2.6338155515370703e-05,
      "loss": 0.0015,
      "step": 2617
    },
    {
      "epoch": 3.31,
      "learning_rate": 2.6329113924050635e-05,
      "loss": 0.1496,
      "step": 2618
    },
    {
      "epoch": 3.31,
      "learning_rate": 2.632007233273056e-05,
      "loss": 0.0033,
      "step": 2619
    },
    {
      "epoch": 3.32,
      "learning_rate": 2.631103074141049e-05,
      "loss": 0.0212,
      "step": 2620
    },
    {
      "epoch": 3.32,
      "learning_rate": 2.6301989150090416e-05,
      "loss": 0.0332,
      "step": 2621
    },
    {
      "epoch": 3.32,
      "learning_rate": 2.629294755877034e-05,
      "loss": 0.0836,
      "step": 2622
    },
    {
      "epoch": 3.32,
      "learning_rate": 2.6283905967450273e-05,
      "loss": 0.0204,
      "step": 2623
    },
    {
      "epoch": 3.32,
      "learning_rate": 2.6274864376130198e-05,
      "loss": 0.004,
      "step": 2624
    },
    {
      "epoch": 3.32,
      "learning_rate": 2.626582278481013e-05,
      "loss": 0.0783,
      "step": 2625
    },
    {
      "epoch": 3.32,
      "learning_rate": 2.6256781193490054e-05,
      "loss": 0.1202,
      "step": 2626
    },
    {
      "epoch": 3.33,
      "learning_rate": 2.6247739602169986e-05,
      "loss": 0.0273,
      "step": 2627
    },
    {
      "epoch": 3.33,
      "learning_rate": 2.623869801084991e-05,
      "loss": 0.0062,
      "step": 2628
    },
    {
      "epoch": 3.33,
      "learning_rate": 2.6229656419529842e-05,
      "loss": 0.0213,
      "step": 2629
    },
    {
      "epoch": 3.33,
      "learning_rate": 2.6220614828209767e-05,
      "loss": 0.0693,
      "step": 2630
    },
    {
      "epoch": 3.33,
      "learning_rate": 2.6211573236889692e-05,
      "loss": 0.0189,
      "step": 2631
    },
    {
      "epoch": 3.33,
      "learning_rate": 2.6202531645569623e-05,
      "loss": 0.0294,
      "step": 2632
    },
    {
      "epoch": 3.33,
      "learning_rate": 2.619349005424955e-05,
      "loss": 0.0013,
      "step": 2633
    },
    {
      "epoch": 3.33,
      "learning_rate": 2.618444846292948e-05,
      "loss": 0.0016,
      "step": 2634
    },
    {
      "epoch": 3.34,
      "learning_rate": 2.6175406871609405e-05,
      "loss": 0.1385,
      "step": 2635
    },
    {
      "epoch": 3.34,
      "learning_rate": 2.6166365280289333e-05,
      "loss": 0.1166,
      "step": 2636
    },
    {
      "epoch": 3.34,
      "learning_rate": 2.615732368896926e-05,
      "loss": 0.1737,
      "step": 2637
    },
    {
      "epoch": 3.34,
      "learning_rate": 2.6148282097649186e-05,
      "loss": 0.0074,
      "step": 2638
    },
    {
      "epoch": 3.34,
      "learning_rate": 2.6139240506329114e-05,
      "loss": 0.0993,
      "step": 2639
    },
    {
      "epoch": 3.34,
      "learning_rate": 2.6130198915009043e-05,
      "loss": 0.0035,
      "step": 2640
    },
    {
      "epoch": 3.34,
      "learning_rate": 2.612115732368897e-05,
      "loss": 0.0128,
      "step": 2641
    },
    {
      "epoch": 3.34,
      "learning_rate": 2.6112115732368896e-05,
      "loss": 0.0015,
      "step": 2642
    },
    {
      "epoch": 3.35,
      "learning_rate": 2.6103074141048827e-05,
      "loss": 0.0451,
      "step": 2643
    },
    {
      "epoch": 3.35,
      "learning_rate": 2.6094032549728752e-05,
      "loss": 0.0502,
      "step": 2644
    },
    {
      "epoch": 3.35,
      "learning_rate": 2.6084990958408684e-05,
      "loss": 0.0359,
      "step": 2645
    },
    {
      "epoch": 3.35,
      "learning_rate": 2.607594936708861e-05,
      "loss": 0.1065,
      "step": 2646
    },
    {
      "epoch": 3.35,
      "learning_rate": 2.6066907775768534e-05,
      "loss": 0.0338,
      "step": 2647
    },
    {
      "epoch": 3.35,
      "learning_rate": 2.6057866184448465e-05,
      "loss": 0.0029,
      "step": 2648
    },
    {
      "epoch": 3.35,
      "learning_rate": 2.604882459312839e-05,
      "loss": 0.0054,
      "step": 2649
    },
    {
      "epoch": 3.35,
      "learning_rate": 2.603978300180832e-05,
      "loss": 0.0144,
      "step": 2650
    },
    {
      "epoch": 3.36,
      "learning_rate": 2.6030741410488246e-05,
      "loss": 0.0015,
      "step": 2651
    },
    {
      "epoch": 3.36,
      "learning_rate": 2.6021699819168178e-05,
      "loss": 0.0411,
      "step": 2652
    },
    {
      "epoch": 3.36,
      "learning_rate": 2.6012658227848103e-05,
      "loss": 0.0022,
      "step": 2653
    },
    {
      "epoch": 3.36,
      "learning_rate": 2.6003616636528028e-05,
      "loss": 0.1876,
      "step": 2654
    },
    {
      "epoch": 3.36,
      "learning_rate": 2.599457504520796e-05,
      "loss": 0.1178,
      "step": 2655
    },
    {
      "epoch": 3.36,
      "learning_rate": 2.5985533453887884e-05,
      "loss": 0.0824,
      "step": 2656
    },
    {
      "epoch": 3.36,
      "learning_rate": 2.5976491862567816e-05,
      "loss": 0.2676,
      "step": 2657
    },
    {
      "epoch": 3.36,
      "learning_rate": 2.596745027124774e-05,
      "loss": 0.0737,
      "step": 2658
    },
    {
      "epoch": 3.37,
      "learning_rate": 2.595840867992767e-05,
      "loss": 0.0737,
      "step": 2659
    },
    {
      "epoch": 3.37,
      "learning_rate": 2.5949367088607597e-05,
      "loss": 0.0089,
      "step": 2660
    },
    {
      "epoch": 3.37,
      "learning_rate": 2.5940325497287525e-05,
      "loss": 0.0069,
      "step": 2661
    },
    {
      "epoch": 3.37,
      "learning_rate": 2.593128390596745e-05,
      "loss": 0.0761,
      "step": 2662
    },
    {
      "epoch": 3.37,
      "learning_rate": 2.592224231464738e-05,
      "loss": 0.0014,
      "step": 2663
    },
    {
      "epoch": 3.37,
      "learning_rate": 2.5913200723327307e-05,
      "loss": 0.1108,
      "step": 2664
    },
    {
      "epoch": 3.37,
      "learning_rate": 2.590415913200723e-05,
      "loss": 0.1585,
      "step": 2665
    },
    {
      "epoch": 3.37,
      "learning_rate": 2.5895117540687163e-05,
      "loss": 0.0033,
      "step": 2666
    },
    {
      "epoch": 3.38,
      "learning_rate": 2.5886075949367088e-05,
      "loss": 0.0025,
      "step": 2667
    },
    {
      "epoch": 3.38,
      "learning_rate": 2.587703435804702e-05,
      "loss": 0.0402,
      "step": 2668
    },
    {
      "epoch": 3.38,
      "learning_rate": 2.5867992766726945e-05,
      "loss": 0.0878,
      "step": 2669
    },
    {
      "epoch": 3.38,
      "learning_rate": 2.585895117540687e-05,
      "loss": 0.1087,
      "step": 2670
    },
    {
      "epoch": 3.38,
      "learning_rate": 2.58499095840868e-05,
      "loss": 0.1001,
      "step": 2671
    },
    {
      "epoch": 3.38,
      "learning_rate": 2.5840867992766726e-05,
      "loss": 0.1208,
      "step": 2672
    },
    {
      "epoch": 3.38,
      "learning_rate": 2.5831826401446658e-05,
      "loss": 0.0053,
      "step": 2673
    },
    {
      "epoch": 3.38,
      "learning_rate": 2.5822784810126582e-05,
      "loss": 0.0133,
      "step": 2674
    },
    {
      "epoch": 3.39,
      "learning_rate": 2.5813743218806514e-05,
      "loss": 0.1453,
      "step": 2675
    },
    {
      "epoch": 3.39,
      "learning_rate": 2.580470162748644e-05,
      "loss": 0.0195,
      "step": 2676
    },
    {
      "epoch": 3.39,
      "learning_rate": 2.579566003616637e-05,
      "loss": 0.0145,
      "step": 2677
    },
    {
      "epoch": 3.39,
      "learning_rate": 2.5786618444846295e-05,
      "loss": 0.0491,
      "step": 2678
    },
    {
      "epoch": 3.39,
      "learning_rate": 2.577757685352622e-05,
      "loss": 0.0029,
      "step": 2679
    },
    {
      "epoch": 3.39,
      "learning_rate": 2.5768535262206152e-05,
      "loss": 0.0667,
      "step": 2680
    },
    {
      "epoch": 3.39,
      "learning_rate": 2.5759493670886077e-05,
      "loss": 0.0168,
      "step": 2681
    },
    {
      "epoch": 3.39,
      "learning_rate": 2.5750452079566005e-05,
      "loss": 0.0764,
      "step": 2682
    },
    {
      "epoch": 3.4,
      "learning_rate": 2.5741410488245933e-05,
      "loss": 0.1302,
      "step": 2683
    },
    {
      "epoch": 3.4,
      "learning_rate": 2.573236889692586e-05,
      "loss": 0.0473,
      "step": 2684
    },
    {
      "epoch": 3.4,
      "learning_rate": 2.5723327305605786e-05,
      "loss": 0.0547,
      "step": 2685
    },
    {
      "epoch": 3.4,
      "learning_rate": 2.5714285714285714e-05,
      "loss": 0.1213,
      "step": 2686
    },
    {
      "epoch": 3.4,
      "learning_rate": 2.5705244122965643e-05,
      "loss": 0.022,
      "step": 2687
    },
    {
      "epoch": 3.4,
      "learning_rate": 2.5696202531645568e-05,
      "loss": 0.0208,
      "step": 2688
    },
    {
      "epoch": 3.4,
      "learning_rate": 2.56871609403255e-05,
      "loss": 0.0171,
      "step": 2689
    },
    {
      "epoch": 3.4,
      "learning_rate": 2.5678119349005424e-05,
      "loss": 0.3485,
      "step": 2690
    },
    {
      "epoch": 3.41,
      "learning_rate": 2.5669077757685356e-05,
      "loss": 0.005,
      "step": 2691
    },
    {
      "epoch": 3.41,
      "learning_rate": 2.566003616636528e-05,
      "loss": 0.1084,
      "step": 2692
    },
    {
      "epoch": 3.41,
      "learning_rate": 2.5650994575045212e-05,
      "loss": 0.0021,
      "step": 2693
    },
    {
      "epoch": 3.41,
      "learning_rate": 2.5641952983725137e-05,
      "loss": 0.1208,
      "step": 2694
    },
    {
      "epoch": 3.41,
      "learning_rate": 2.5632911392405062e-05,
      "loss": 0.0572,
      "step": 2695
    },
    {
      "epoch": 3.41,
      "learning_rate": 2.5623869801084993e-05,
      "loss": 0.0037,
      "step": 2696
    },
    {
      "epoch": 3.41,
      "learning_rate": 2.5614828209764918e-05,
      "loss": 0.035,
      "step": 2697
    },
    {
      "epoch": 3.41,
      "learning_rate": 2.560578661844485e-05,
      "loss": 0.2419,
      "step": 2698
    },
    {
      "epoch": 3.42,
      "learning_rate": 2.5596745027124775e-05,
      "loss": 0.024,
      "step": 2699
    },
    {
      "epoch": 3.42,
      "learning_rate": 2.5587703435804706e-05,
      "loss": 0.0278,
      "step": 2700
    },
    {
      "epoch": 3.42,
      "learning_rate": 2.557866184448463e-05,
      "loss": 0.0075,
      "step": 2701
    },
    {
      "epoch": 3.42,
      "learning_rate": 2.5569620253164556e-05,
      "loss": 0.0345,
      "step": 2702
    },
    {
      "epoch": 3.42,
      "learning_rate": 2.5560578661844488e-05,
      "loss": 0.0095,
      "step": 2703
    },
    {
      "epoch": 3.42,
      "learning_rate": 2.5551537070524413e-05,
      "loss": 0.0061,
      "step": 2704
    },
    {
      "epoch": 3.42,
      "learning_rate": 2.554249547920434e-05,
      "loss": 0.4344,
      "step": 2705
    },
    {
      "epoch": 3.43,
      "learning_rate": 2.553345388788427e-05,
      "loss": 0.0822,
      "step": 2706
    },
    {
      "epoch": 3.43,
      "learning_rate": 2.5524412296564197e-05,
      "loss": 0.1959,
      "step": 2707
    },
    {
      "epoch": 3.43,
      "learning_rate": 2.5515370705244122e-05,
      "loss": 0.0064,
      "step": 2708
    },
    {
      "epoch": 3.43,
      "learning_rate": 2.5506329113924054e-05,
      "loss": 0.0086,
      "step": 2709
    },
    {
      "epoch": 3.43,
      "learning_rate": 2.549728752260398e-05,
      "loss": 0.1046,
      "step": 2710
    },
    {
      "epoch": 3.43,
      "learning_rate": 2.5488245931283903e-05,
      "loss": 0.0036,
      "step": 2711
    },
    {
      "epoch": 3.43,
      "learning_rate": 2.5479204339963835e-05,
      "loss": 0.0092,
      "step": 2712
    },
    {
      "epoch": 3.43,
      "learning_rate": 2.547016274864376e-05,
      "loss": 0.2012,
      "step": 2713
    },
    {
      "epoch": 3.44,
      "learning_rate": 2.546112115732369e-05,
      "loss": 0.0038,
      "step": 2714
    },
    {
      "epoch": 3.44,
      "learning_rate": 2.5452079566003616e-05,
      "loss": 0.0108,
      "step": 2715
    },
    {
      "epoch": 3.44,
      "learning_rate": 2.5443037974683548e-05,
      "loss": 0.1695,
      "step": 2716
    },
    {
      "epoch": 3.44,
      "learning_rate": 2.5433996383363473e-05,
      "loss": 0.0113,
      "step": 2717
    },
    {
      "epoch": 3.44,
      "learning_rate": 2.5424954792043398e-05,
      "loss": 0.0021,
      "step": 2718
    },
    {
      "epoch": 3.44,
      "learning_rate": 2.541591320072333e-05,
      "loss": 0.0035,
      "step": 2719
    },
    {
      "epoch": 3.44,
      "learning_rate": 2.5406871609403254e-05,
      "loss": 0.101,
      "step": 2720
    },
    {
      "epoch": 3.44,
      "learning_rate": 2.5397830018083186e-05,
      "loss": 0.015,
      "step": 2721
    },
    {
      "epoch": 3.45,
      "learning_rate": 2.538878842676311e-05,
      "loss": 0.0142,
      "step": 2722
    },
    {
      "epoch": 3.45,
      "learning_rate": 2.5379746835443042e-05,
      "loss": 0.0133,
      "step": 2723
    },
    {
      "epoch": 3.45,
      "learning_rate": 2.5370705244122967e-05,
      "loss": 0.0055,
      "step": 2724
    },
    {
      "epoch": 3.45,
      "learning_rate": 2.53616636528029e-05,
      "loss": 0.0563,
      "step": 2725
    },
    {
      "epoch": 3.45,
      "learning_rate": 2.5352622061482824e-05,
      "loss": 0.0148,
      "step": 2726
    },
    {
      "epoch": 3.45,
      "learning_rate": 2.534358047016275e-05,
      "loss": 0.0494,
      "step": 2727
    },
    {
      "epoch": 3.45,
      "learning_rate": 2.533453887884268e-05,
      "loss": 0.0067,
      "step": 2728
    },
    {
      "epoch": 3.45,
      "learning_rate": 2.5325497287522605e-05,
      "loss": 0.0037,
      "step": 2729
    },
    {
      "epoch": 3.46,
      "learning_rate": 2.5316455696202533e-05,
      "loss": 0.0049,
      "step": 2730
    },
    {
      "epoch": 3.46,
      "learning_rate": 2.530741410488246e-05,
      "loss": 0.0021,
      "step": 2731
    },
    {
      "epoch": 3.46,
      "learning_rate": 2.529837251356239e-05,
      "loss": 0.0887,
      "step": 2732
    },
    {
      "epoch": 3.46,
      "learning_rate": 2.5289330922242315e-05,
      "loss": 0.3909,
      "step": 2733
    },
    {
      "epoch": 3.46,
      "learning_rate": 2.528028933092224e-05,
      "loss": 0.1276,
      "step": 2734
    },
    {
      "epoch": 3.46,
      "learning_rate": 2.527124773960217e-05,
      "loss": 0.0066,
      "step": 2735
    },
    {
      "epoch": 3.46,
      "learning_rate": 2.5262206148282096e-05,
      "loss": 0.0229,
      "step": 2736
    },
    {
      "epoch": 3.46,
      "learning_rate": 2.5253164556962027e-05,
      "loss": 0.124,
      "step": 2737
    },
    {
      "epoch": 3.47,
      "learning_rate": 2.5244122965641952e-05,
      "loss": 0.0293,
      "step": 2738
    },
    {
      "epoch": 3.47,
      "learning_rate": 2.5235081374321884e-05,
      "loss": 0.0105,
      "step": 2739
    },
    {
      "epoch": 3.47,
      "learning_rate": 2.522603978300181e-05,
      "loss": 0.1019,
      "step": 2740
    },
    {
      "epoch": 3.47,
      "learning_rate": 2.521699819168174e-05,
      "loss": 0.2998,
      "step": 2741
    },
    {
      "epoch": 3.47,
      "learning_rate": 2.5207956600361665e-05,
      "loss": 0.0155,
      "step": 2742
    },
    {
      "epoch": 3.47,
      "learning_rate": 2.519891500904159e-05,
      "loss": 0.0274,
      "step": 2743
    },
    {
      "epoch": 3.47,
      "learning_rate": 2.5189873417721522e-05,
      "loss": 0.0127,
      "step": 2744
    },
    {
      "epoch": 3.47,
      "learning_rate": 2.5180831826401447e-05,
      "loss": 0.0371,
      "step": 2745
    },
    {
      "epoch": 3.48,
      "learning_rate": 2.5171790235081378e-05,
      "loss": 0.0121,
      "step": 2746
    },
    {
      "epoch": 3.48,
      "learning_rate": 2.5162748643761303e-05,
      "loss": 0.0087,
      "step": 2747
    },
    {
      "epoch": 3.48,
      "learning_rate": 2.5153707052441235e-05,
      "loss": 0.0116,
      "step": 2748
    },
    {
      "epoch": 3.48,
      "learning_rate": 2.514466546112116e-05,
      "loss": 0.0118,
      "step": 2749
    },
    {
      "epoch": 3.48,
      "learning_rate": 2.5135623869801084e-05,
      "loss": 0.2103,
      "step": 2750
    },
    {
      "epoch": 3.48,
      "eval_accuracy": 0.9031531531531531,
      "eval_loss": 0.39241811633110046,
      "eval_runtime": 45.1121,
      "eval_samples_per_second": 29.526,
      "eval_steps_per_second": 3.702,
      "step": 2750
    },
    {
      "epoch": 3.48,
      "learning_rate": 2.5126582278481016e-05,
      "loss": 0.0433,
      "step": 2751
    },
    {
      "epoch": 3.48,
      "learning_rate": 2.511754068716094e-05,
      "loss": 0.0122,
      "step": 2752
    },
    {
      "epoch": 3.48,
      "learning_rate": 2.510849909584087e-05,
      "loss": 0.2189,
      "step": 2753
    },
    {
      "epoch": 3.49,
      "learning_rate": 2.5099457504520797e-05,
      "loss": 0.098,
      "step": 2754
    },
    {
      "epoch": 3.49,
      "learning_rate": 2.5090415913200726e-05,
      "loss": 0.0068,
      "step": 2755
    },
    {
      "epoch": 3.49,
      "learning_rate": 2.508137432188065e-05,
      "loss": 0.0225,
      "step": 2756
    },
    {
      "epoch": 3.49,
      "learning_rate": 2.5072332730560582e-05,
      "loss": 0.0144,
      "step": 2757
    },
    {
      "epoch": 3.49,
      "learning_rate": 2.5063291139240507e-05,
      "loss": 0.0472,
      "step": 2758
    },
    {
      "epoch": 3.49,
      "learning_rate": 2.5054249547920432e-05,
      "loss": 0.0904,
      "step": 2759
    },
    {
      "epoch": 3.49,
      "learning_rate": 2.5045207956600363e-05,
      "loss": 0.176,
      "step": 2760
    },
    {
      "epoch": 3.49,
      "learning_rate": 2.5036166365280288e-05,
      "loss": 0.0214,
      "step": 2761
    },
    {
      "epoch": 3.5,
      "learning_rate": 2.502712477396022e-05,
      "loss": 0.0539,
      "step": 2762
    },
    {
      "epoch": 3.5,
      "learning_rate": 2.5018083182640145e-05,
      "loss": 0.0151,
      "step": 2763
    },
    {
      "epoch": 3.5,
      "learning_rate": 2.5009041591320076e-05,
      "loss": 0.0047,
      "step": 2764
    },
    {
      "epoch": 3.5,
      "learning_rate": 2.5e-05,
      "loss": 0.0584,
      "step": 2765
    },
    {
      "epoch": 3.5,
      "learning_rate": 2.499095840867993e-05,
      "loss": 0.0073,
      "step": 2766
    },
    {
      "epoch": 3.5,
      "learning_rate": 2.4981916817359858e-05,
      "loss": 0.0076,
      "step": 2767
    },
    {
      "epoch": 3.5,
      "learning_rate": 2.4972875226039786e-05,
      "loss": 0.336,
      "step": 2768
    },
    {
      "epoch": 3.5,
      "learning_rate": 2.496383363471971e-05,
      "loss": 0.0105,
      "step": 2769
    },
    {
      "epoch": 3.51,
      "learning_rate": 2.495479204339964e-05,
      "loss": 0.0122,
      "step": 2770
    },
    {
      "epoch": 3.51,
      "learning_rate": 2.4945750452079567e-05,
      "loss": 0.0047,
      "step": 2771
    },
    {
      "epoch": 3.51,
      "learning_rate": 2.4936708860759495e-05,
      "loss": 0.0066,
      "step": 2772
    },
    {
      "epoch": 3.51,
      "learning_rate": 2.4927667269439424e-05,
      "loss": 0.0954,
      "step": 2773
    },
    {
      "epoch": 3.51,
      "learning_rate": 2.4918625678119352e-05,
      "loss": 0.0035,
      "step": 2774
    },
    {
      "epoch": 3.51,
      "learning_rate": 2.490958408679928e-05,
      "loss": 0.0353,
      "step": 2775
    },
    {
      "epoch": 3.51,
      "learning_rate": 2.4900542495479205e-05,
      "loss": 0.0344,
      "step": 2776
    },
    {
      "epoch": 3.51,
      "learning_rate": 2.4891500904159133e-05,
      "loss": 0.1968,
      "step": 2777
    },
    {
      "epoch": 3.52,
      "learning_rate": 2.488245931283906e-05,
      "loss": 0.0837,
      "step": 2778
    },
    {
      "epoch": 3.52,
      "learning_rate": 2.4873417721518986e-05,
      "loss": 0.0041,
      "step": 2779
    },
    {
      "epoch": 3.52,
      "learning_rate": 2.4864376130198915e-05,
      "loss": 0.0098,
      "step": 2780
    },
    {
      "epoch": 3.52,
      "learning_rate": 2.4855334538878843e-05,
      "loss": 0.0099,
      "step": 2781
    },
    {
      "epoch": 3.52,
      "learning_rate": 2.484629294755877e-05,
      "loss": 0.0271,
      "step": 2782
    },
    {
      "epoch": 3.52,
      "learning_rate": 2.48372513562387e-05,
      "loss": 0.1436,
      "step": 2783
    },
    {
      "epoch": 3.52,
      "learning_rate": 2.4828209764918628e-05,
      "loss": 0.1997,
      "step": 2784
    },
    {
      "epoch": 3.52,
      "learning_rate": 2.4819168173598552e-05,
      "loss": 0.0308,
      "step": 2785
    },
    {
      "epoch": 3.53,
      "learning_rate": 2.481012658227848e-05,
      "loss": 0.2002,
      "step": 2786
    },
    {
      "epoch": 3.53,
      "learning_rate": 2.480108499095841e-05,
      "loss": 0.0679,
      "step": 2787
    },
    {
      "epoch": 3.53,
      "learning_rate": 2.4792043399638337e-05,
      "loss": 0.0238,
      "step": 2788
    },
    {
      "epoch": 3.53,
      "learning_rate": 2.4783001808318265e-05,
      "loss": 0.0036,
      "step": 2789
    },
    {
      "epoch": 3.53,
      "learning_rate": 2.4773960216998194e-05,
      "loss": 0.2046,
      "step": 2790
    },
    {
      "epoch": 3.53,
      "learning_rate": 2.4764918625678122e-05,
      "loss": 0.0088,
      "step": 2791
    },
    {
      "epoch": 3.53,
      "learning_rate": 2.475587703435805e-05,
      "loss": 0.0322,
      "step": 2792
    },
    {
      "epoch": 3.54,
      "learning_rate": 2.4746835443037975e-05,
      "loss": 0.0031,
      "step": 2793
    },
    {
      "epoch": 3.54,
      "learning_rate": 2.4737793851717903e-05,
      "loss": 0.0155,
      "step": 2794
    },
    {
      "epoch": 3.54,
      "learning_rate": 2.472875226039783e-05,
      "loss": 0.0031,
      "step": 2795
    },
    {
      "epoch": 3.54,
      "learning_rate": 2.471971066907776e-05,
      "loss": 0.192,
      "step": 2796
    },
    {
      "epoch": 3.54,
      "learning_rate": 2.4710669077757688e-05,
      "loss": 0.1921,
      "step": 2797
    },
    {
      "epoch": 3.54,
      "learning_rate": 2.4701627486437616e-05,
      "loss": 0.0971,
      "step": 2798
    },
    {
      "epoch": 3.54,
      "learning_rate": 2.469258589511754e-05,
      "loss": 0.191,
      "step": 2799
    },
    {
      "epoch": 3.54,
      "learning_rate": 2.468354430379747e-05,
      "loss": 0.0513,
      "step": 2800
    },
    {
      "epoch": 3.55,
      "learning_rate": 2.4674502712477397e-05,
      "loss": 0.0193,
      "step": 2801
    },
    {
      "epoch": 3.55,
      "learning_rate": 2.4665461121157322e-05,
      "loss": 0.0292,
      "step": 2802
    },
    {
      "epoch": 3.55,
      "learning_rate": 2.465641952983725e-05,
      "loss": 0.0882,
      "step": 2803
    },
    {
      "epoch": 3.55,
      "learning_rate": 2.464737793851718e-05,
      "loss": 0.1259,
      "step": 2804
    },
    {
      "epoch": 3.55,
      "learning_rate": 2.4638336347197107e-05,
      "loss": 0.006,
      "step": 2805
    },
    {
      "epoch": 3.55,
      "learning_rate": 2.4629294755877035e-05,
      "loss": 0.0062,
      "step": 2806
    },
    {
      "epoch": 3.55,
      "learning_rate": 2.4620253164556963e-05,
      "loss": 0.006,
      "step": 2807
    },
    {
      "epoch": 3.55,
      "learning_rate": 2.4611211573236892e-05,
      "loss": 0.0062,
      "step": 2808
    },
    {
      "epoch": 3.56,
      "learning_rate": 2.4602169981916817e-05,
      "loss": 0.0078,
      "step": 2809
    },
    {
      "epoch": 3.56,
      "learning_rate": 2.4593128390596745e-05,
      "loss": 0.0353,
      "step": 2810
    },
    {
      "epoch": 3.56,
      "learning_rate": 2.4584086799276673e-05,
      "loss": 0.2469,
      "step": 2811
    },
    {
      "epoch": 3.56,
      "learning_rate": 2.45750452079566e-05,
      "loss": 0.0046,
      "step": 2812
    },
    {
      "epoch": 3.56,
      "learning_rate": 2.456600361663653e-05,
      "loss": 0.0027,
      "step": 2813
    },
    {
      "epoch": 3.56,
      "learning_rate": 2.4556962025316458e-05,
      "loss": 0.0357,
      "step": 2814
    },
    {
      "epoch": 3.56,
      "learning_rate": 2.4547920433996386e-05,
      "loss": 0.0149,
      "step": 2815
    },
    {
      "epoch": 3.56,
      "learning_rate": 2.4538878842676314e-05,
      "loss": 0.1389,
      "step": 2816
    },
    {
      "epoch": 3.57,
      "learning_rate": 2.452983725135624e-05,
      "loss": 0.0621,
      "step": 2817
    },
    {
      "epoch": 3.57,
      "learning_rate": 2.4520795660036167e-05,
      "loss": 0.1036,
      "step": 2818
    },
    {
      "epoch": 3.57,
      "learning_rate": 2.4511754068716096e-05,
      "loss": 0.1411,
      "step": 2819
    },
    {
      "epoch": 3.57,
      "learning_rate": 2.4502712477396024e-05,
      "loss": 0.1489,
      "step": 2820
    },
    {
      "epoch": 3.57,
      "learning_rate": 2.4493670886075952e-05,
      "loss": 0.0152,
      "step": 2821
    },
    {
      "epoch": 3.57,
      "learning_rate": 2.448462929475588e-05,
      "loss": 0.0042,
      "step": 2822
    },
    {
      "epoch": 3.57,
      "learning_rate": 2.4475587703435805e-05,
      "loss": 0.0031,
      "step": 2823
    },
    {
      "epoch": 3.57,
      "learning_rate": 2.4466546112115733e-05,
      "loss": 0.0037,
      "step": 2824
    },
    {
      "epoch": 3.58,
      "learning_rate": 2.445750452079566e-05,
      "loss": 0.0025,
      "step": 2825
    },
    {
      "epoch": 3.58,
      "learning_rate": 2.4448462929475586e-05,
      "loss": 0.0137,
      "step": 2826
    },
    {
      "epoch": 3.58,
      "learning_rate": 2.4439421338155515e-05,
      "loss": 0.0079,
      "step": 2827
    },
    {
      "epoch": 3.58,
      "learning_rate": 2.4430379746835443e-05,
      "loss": 0.0359,
      "step": 2828
    },
    {
      "epoch": 3.58,
      "learning_rate": 2.442133815551537e-05,
      "loss": 0.127,
      "step": 2829
    },
    {
      "epoch": 3.58,
      "learning_rate": 2.44122965641953e-05,
      "loss": 0.1521,
      "step": 2830
    },
    {
      "epoch": 3.58,
      "learning_rate": 2.4403254972875228e-05,
      "loss": 0.0952,
      "step": 2831
    },
    {
      "epoch": 3.58,
      "learning_rate": 2.4394213381555156e-05,
      "loss": 0.0337,
      "step": 2832
    },
    {
      "epoch": 3.59,
      "learning_rate": 2.438517179023508e-05,
      "loss": 0.1041,
      "step": 2833
    },
    {
      "epoch": 3.59,
      "learning_rate": 2.437613019891501e-05,
      "loss": 0.0912,
      "step": 2834
    },
    {
      "epoch": 3.59,
      "learning_rate": 2.4367088607594937e-05,
      "loss": 0.0016,
      "step": 2835
    },
    {
      "epoch": 3.59,
      "learning_rate": 2.4358047016274865e-05,
      "loss": 0.0342,
      "step": 2836
    },
    {
      "epoch": 3.59,
      "learning_rate": 2.4349005424954794e-05,
      "loss": 0.0039,
      "step": 2837
    },
    {
      "epoch": 3.59,
      "learning_rate": 2.4339963833634722e-05,
      "loss": 0.0064,
      "step": 2838
    },
    {
      "epoch": 3.59,
      "learning_rate": 2.433092224231465e-05,
      "loss": 0.003,
      "step": 2839
    },
    {
      "epoch": 3.59,
      "learning_rate": 2.432188065099458e-05,
      "loss": 0.091,
      "step": 2840
    },
    {
      "epoch": 3.6,
      "learning_rate": 2.4312839059674503e-05,
      "loss": 0.0383,
      "step": 2841
    },
    {
      "epoch": 3.6,
      "learning_rate": 2.430379746835443e-05,
      "loss": 0.0072,
      "step": 2842
    },
    {
      "epoch": 3.6,
      "learning_rate": 2.429475587703436e-05,
      "loss": 0.141,
      "step": 2843
    },
    {
      "epoch": 3.6,
      "learning_rate": 2.4285714285714288e-05,
      "loss": 0.0018,
      "step": 2844
    },
    {
      "epoch": 3.6,
      "learning_rate": 2.4276672694394216e-05,
      "loss": 0.0036,
      "step": 2845
    },
    {
      "epoch": 3.6,
      "learning_rate": 2.426763110307414e-05,
      "loss": 0.094,
      "step": 2846
    },
    {
      "epoch": 3.6,
      "learning_rate": 2.425858951175407e-05,
      "loss": 0.1063,
      "step": 2847
    },
    {
      "epoch": 3.6,
      "learning_rate": 2.4249547920433998e-05,
      "loss": 0.0181,
      "step": 2848
    },
    {
      "epoch": 3.61,
      "learning_rate": 2.4240506329113922e-05,
      "loss": 0.0512,
      "step": 2849
    },
    {
      "epoch": 3.61,
      "learning_rate": 2.423146473779385e-05,
      "loss": 0.0213,
      "step": 2850
    },
    {
      "epoch": 3.61,
      "learning_rate": 2.422242314647378e-05,
      "loss": 0.0135,
      "step": 2851
    },
    {
      "epoch": 3.61,
      "learning_rate": 2.4213381555153707e-05,
      "loss": 0.0095,
      "step": 2852
    },
    {
      "epoch": 3.61,
      "learning_rate": 2.4204339963833635e-05,
      "loss": 0.03,
      "step": 2853
    },
    {
      "epoch": 3.61,
      "learning_rate": 2.4195298372513564e-05,
      "loss": 0.0171,
      "step": 2854
    },
    {
      "epoch": 3.61,
      "learning_rate": 2.4186256781193492e-05,
      "loss": 0.2265,
      "step": 2855
    },
    {
      "epoch": 3.61,
      "learning_rate": 2.417721518987342e-05,
      "loss": 0.0027,
      "step": 2856
    },
    {
      "epoch": 3.62,
      "learning_rate": 2.4168173598553345e-05,
      "loss": 0.0065,
      "step": 2857
    },
    {
      "epoch": 3.62,
      "learning_rate": 2.4159132007233273e-05,
      "loss": 0.256,
      "step": 2858
    },
    {
      "epoch": 3.62,
      "learning_rate": 2.41500904159132e-05,
      "loss": 0.0086,
      "step": 2859
    },
    {
      "epoch": 3.62,
      "learning_rate": 2.414104882459313e-05,
      "loss": 0.0023,
      "step": 2860
    },
    {
      "epoch": 3.62,
      "learning_rate": 2.4132007233273058e-05,
      "loss": 0.0028,
      "step": 2861
    },
    {
      "epoch": 3.62,
      "learning_rate": 2.4122965641952986e-05,
      "loss": 0.1745,
      "step": 2862
    },
    {
      "epoch": 3.62,
      "learning_rate": 2.4113924050632914e-05,
      "loss": 0.0023,
      "step": 2863
    },
    {
      "epoch": 3.62,
      "learning_rate": 2.4104882459312843e-05,
      "loss": 0.0161,
      "step": 2864
    },
    {
      "epoch": 3.63,
      "learning_rate": 2.4095840867992767e-05,
      "loss": 0.006,
      "step": 2865
    },
    {
      "epoch": 3.63,
      "learning_rate": 2.4086799276672696e-05,
      "loss": 0.173,
      "step": 2866
    },
    {
      "epoch": 3.63,
      "learning_rate": 2.4077757685352624e-05,
      "loss": 0.0513,
      "step": 2867
    },
    {
      "epoch": 3.63,
      "learning_rate": 2.4068716094032552e-05,
      "loss": 0.0075,
      "step": 2868
    },
    {
      "epoch": 3.63,
      "learning_rate": 2.405967450271248e-05,
      "loss": 0.1578,
      "step": 2869
    },
    {
      "epoch": 3.63,
      "learning_rate": 2.4050632911392405e-05,
      "loss": 0.0088,
      "step": 2870
    },
    {
      "epoch": 3.63,
      "learning_rate": 2.4041591320072333e-05,
      "loss": 0.0012,
      "step": 2871
    },
    {
      "epoch": 3.64,
      "learning_rate": 2.403254972875226e-05,
      "loss": 0.006,
      "step": 2872
    },
    {
      "epoch": 3.64,
      "learning_rate": 2.4023508137432187e-05,
      "loss": 0.0081,
      "step": 2873
    },
    {
      "epoch": 3.64,
      "learning_rate": 2.4014466546112115e-05,
      "loss": 0.0025,
      "step": 2874
    },
    {
      "epoch": 3.64,
      "learning_rate": 2.4005424954792043e-05,
      "loss": 0.0045,
      "step": 2875
    },
    {
      "epoch": 3.64,
      "learning_rate": 2.399638336347197e-05,
      "loss": 0.0378,
      "step": 2876
    },
    {
      "epoch": 3.64,
      "learning_rate": 2.39873417721519e-05,
      "loss": 0.0064,
      "step": 2877
    },
    {
      "epoch": 3.64,
      "learning_rate": 2.3978300180831828e-05,
      "loss": 0.0084,
      "step": 2878
    },
    {
      "epoch": 3.64,
      "learning_rate": 2.3969258589511756e-05,
      "loss": 0.0795,
      "step": 2879
    },
    {
      "epoch": 3.65,
      "learning_rate": 2.3960216998191684e-05,
      "loss": 0.029,
      "step": 2880
    },
    {
      "epoch": 3.65,
      "learning_rate": 2.395117540687161e-05,
      "loss": 0.0038,
      "step": 2881
    },
    {
      "epoch": 3.65,
      "learning_rate": 2.3942133815551537e-05,
      "loss": 0.0109,
      "step": 2882
    },
    {
      "epoch": 3.65,
      "learning_rate": 2.3933092224231466e-05,
      "loss": 0.0022,
      "step": 2883
    },
    {
      "epoch": 3.65,
      "learning_rate": 2.3924050632911394e-05,
      "loss": 0.0058,
      "step": 2884
    },
    {
      "epoch": 3.65,
      "learning_rate": 2.3915009041591322e-05,
      "loss": 0.1352,
      "step": 2885
    },
    {
      "epoch": 3.65,
      "learning_rate": 2.390596745027125e-05,
      "loss": 0.0769,
      "step": 2886
    },
    {
      "epoch": 3.65,
      "learning_rate": 2.389692585895118e-05,
      "loss": 0.0095,
      "step": 2887
    },
    {
      "epoch": 3.66,
      "learning_rate": 2.3887884267631107e-05,
      "loss": 0.017,
      "step": 2888
    },
    {
      "epoch": 3.66,
      "learning_rate": 2.387884267631103e-05,
      "loss": 0.0285,
      "step": 2889
    },
    {
      "epoch": 3.66,
      "learning_rate": 2.386980108499096e-05,
      "loss": 0.0768,
      "step": 2890
    },
    {
      "epoch": 3.66,
      "learning_rate": 2.3860759493670888e-05,
      "loss": 0.1906,
      "step": 2891
    },
    {
      "epoch": 3.66,
      "learning_rate": 2.3851717902350816e-05,
      "loss": 0.0012,
      "step": 2892
    },
    {
      "epoch": 3.66,
      "learning_rate": 2.384267631103074e-05,
      "loss": 0.2828,
      "step": 2893
    },
    {
      "epoch": 3.66,
      "learning_rate": 2.383363471971067e-05,
      "loss": 0.0117,
      "step": 2894
    },
    {
      "epoch": 3.66,
      "learning_rate": 2.3824593128390598e-05,
      "loss": 0.0028,
      "step": 2895
    },
    {
      "epoch": 3.67,
      "learning_rate": 2.3815551537070526e-05,
      "loss": 0.0061,
      "step": 2896
    },
    {
      "epoch": 3.67,
      "learning_rate": 2.380650994575045e-05,
      "loss": 0.0022,
      "step": 2897
    },
    {
      "epoch": 3.67,
      "learning_rate": 2.379746835443038e-05,
      "loss": 0.0065,
      "step": 2898
    },
    {
      "epoch": 3.67,
      "learning_rate": 2.3788426763110307e-05,
      "loss": 0.0761,
      "step": 2899
    },
    {
      "epoch": 3.67,
      "learning_rate": 2.3779385171790235e-05,
      "loss": 0.1419,
      "step": 2900
    },
    {
      "epoch": 3.67,
      "learning_rate": 2.3770343580470164e-05,
      "loss": 0.4393,
      "step": 2901
    },
    {
      "epoch": 3.67,
      "learning_rate": 2.3761301989150092e-05,
      "loss": 0.0138,
      "step": 2902
    },
    {
      "epoch": 3.67,
      "learning_rate": 2.375226039783002e-05,
      "loss": 0.0057,
      "step": 2903
    },
    {
      "epoch": 3.68,
      "learning_rate": 2.374321880650995e-05,
      "loss": 0.1366,
      "step": 2904
    },
    {
      "epoch": 3.68,
      "learning_rate": 2.3734177215189873e-05,
      "loss": 0.0367,
      "step": 2905
    },
    {
      "epoch": 3.68,
      "learning_rate": 2.37251356238698e-05,
      "loss": 0.0377,
      "step": 2906
    },
    {
      "epoch": 3.68,
      "learning_rate": 2.371609403254973e-05,
      "loss": 0.0257,
      "step": 2907
    },
    {
      "epoch": 3.68,
      "learning_rate": 2.3707052441229658e-05,
      "loss": 0.0202,
      "step": 2908
    },
    {
      "epoch": 3.68,
      "learning_rate": 2.3698010849909586e-05,
      "loss": 0.0009,
      "step": 2909
    },
    {
      "epoch": 3.68,
      "learning_rate": 2.3688969258589514e-05,
      "loss": 0.0093,
      "step": 2910
    },
    {
      "epoch": 3.68,
      "learning_rate": 2.3679927667269443e-05,
      "loss": 0.0023,
      "step": 2911
    },
    {
      "epoch": 3.69,
      "learning_rate": 2.367088607594937e-05,
      "loss": 0.1646,
      "step": 2912
    },
    {
      "epoch": 3.69,
      "learning_rate": 2.3661844484629296e-05,
      "loss": 0.0048,
      "step": 2913
    },
    {
      "epoch": 3.69,
      "learning_rate": 2.3652802893309224e-05,
      "loss": 0.0096,
      "step": 2914
    },
    {
      "epoch": 3.69,
      "learning_rate": 2.3643761301989152e-05,
      "loss": 0.0073,
      "step": 2915
    },
    {
      "epoch": 3.69,
      "learning_rate": 2.363471971066908e-05,
      "loss": 0.2111,
      "step": 2916
    },
    {
      "epoch": 3.69,
      "learning_rate": 2.3625678119349005e-05,
      "loss": 0.0023,
      "step": 2917
    },
    {
      "epoch": 3.69,
      "learning_rate": 2.3616636528028933e-05,
      "loss": 0.0045,
      "step": 2918
    },
    {
      "epoch": 3.69,
      "learning_rate": 2.3607594936708862e-05,
      "loss": 0.0043,
      "step": 2919
    },
    {
      "epoch": 3.7,
      "learning_rate": 2.359855334538879e-05,
      "loss": 0.2429,
      "step": 2920
    },
    {
      "epoch": 3.7,
      "learning_rate": 2.3589511754068715e-05,
      "loss": 0.0016,
      "step": 2921
    },
    {
      "epoch": 3.7,
      "learning_rate": 2.3580470162748643e-05,
      "loss": 0.1196,
      "step": 2922
    },
    {
      "epoch": 3.7,
      "learning_rate": 2.357142857142857e-05,
      "loss": 0.0095,
      "step": 2923
    },
    {
      "epoch": 3.7,
      "learning_rate": 2.35623869801085e-05,
      "loss": 0.0331,
      "step": 2924
    },
    {
      "epoch": 3.7,
      "learning_rate": 2.3553345388788428e-05,
      "loss": 0.0055,
      "step": 2925
    },
    {
      "epoch": 3.7,
      "learning_rate": 2.3544303797468356e-05,
      "loss": 0.0511,
      "step": 2926
    },
    {
      "epoch": 3.7,
      "learning_rate": 2.3535262206148284e-05,
      "loss": 0.0567,
      "step": 2927
    },
    {
      "epoch": 3.71,
      "learning_rate": 2.3526220614828212e-05,
      "loss": 0.002,
      "step": 2928
    },
    {
      "epoch": 3.71,
      "learning_rate": 2.3517179023508137e-05,
      "loss": 0.0773,
      "step": 2929
    },
    {
      "epoch": 3.71,
      "learning_rate": 2.3508137432188066e-05,
      "loss": 0.009,
      "step": 2930
    },
    {
      "epoch": 3.71,
      "learning_rate": 2.3499095840867994e-05,
      "loss": 0.0198,
      "step": 2931
    },
    {
      "epoch": 3.71,
      "learning_rate": 2.3490054249547922e-05,
      "loss": 0.0013,
      "step": 2932
    },
    {
      "epoch": 3.71,
      "learning_rate": 2.348101265822785e-05,
      "loss": 0.0812,
      "step": 2933
    },
    {
      "epoch": 3.71,
      "learning_rate": 2.347197106690778e-05,
      "loss": 0.1482,
      "step": 2934
    },
    {
      "epoch": 3.71,
      "learning_rate": 2.3462929475587707e-05,
      "loss": 0.0038,
      "step": 2935
    },
    {
      "epoch": 3.72,
      "learning_rate": 2.3453887884267635e-05,
      "loss": 0.1862,
      "step": 2936
    },
    {
      "epoch": 3.72,
      "learning_rate": 2.344484629294756e-05,
      "loss": 0.039,
      "step": 2937
    },
    {
      "epoch": 3.72,
      "learning_rate": 2.3435804701627488e-05,
      "loss": 0.0038,
      "step": 2938
    },
    {
      "epoch": 3.72,
      "learning_rate": 2.3426763110307416e-05,
      "loss": 0.0077,
      "step": 2939
    },
    {
      "epoch": 3.72,
      "learning_rate": 2.341772151898734e-05,
      "loss": 0.0129,
      "step": 2940
    },
    {
      "epoch": 3.72,
      "learning_rate": 2.340867992766727e-05,
      "loss": 0.0112,
      "step": 2941
    },
    {
      "epoch": 3.72,
      "learning_rate": 2.3399638336347198e-05,
      "loss": 0.0098,
      "step": 2942
    },
    {
      "epoch": 3.72,
      "learning_rate": 2.3390596745027126e-05,
      "loss": 0.0463,
      "step": 2943
    },
    {
      "epoch": 3.73,
      "learning_rate": 2.3381555153707054e-05,
      "loss": 0.0347,
      "step": 2944
    },
    {
      "epoch": 3.73,
      "learning_rate": 2.337251356238698e-05,
      "loss": 0.0033,
      "step": 2945
    },
    {
      "epoch": 3.73,
      "learning_rate": 2.3363471971066907e-05,
      "loss": 0.1538,
      "step": 2946
    },
    {
      "epoch": 3.73,
      "learning_rate": 2.3354430379746835e-05,
      "loss": 0.1951,
      "step": 2947
    },
    {
      "epoch": 3.73,
      "learning_rate": 2.3345388788426764e-05,
      "loss": 0.0163,
      "step": 2948
    },
    {
      "epoch": 3.73,
      "learning_rate": 2.3336347197106692e-05,
      "loss": 0.0043,
      "step": 2949
    },
    {
      "epoch": 3.73,
      "learning_rate": 2.332730560578662e-05,
      "loss": 0.0008,
      "step": 2950
    },
    {
      "epoch": 3.73,
      "learning_rate": 2.331826401446655e-05,
      "loss": 0.0027,
      "step": 2951
    },
    {
      "epoch": 3.74,
      "learning_rate": 2.3309222423146477e-05,
      "loss": 0.0345,
      "step": 2952
    },
    {
      "epoch": 3.74,
      "learning_rate": 2.33001808318264e-05,
      "loss": 0.005,
      "step": 2953
    },
    {
      "epoch": 3.74,
      "learning_rate": 2.329113924050633e-05,
      "loss": 0.101,
      "step": 2954
    },
    {
      "epoch": 3.74,
      "learning_rate": 2.3282097649186258e-05,
      "loss": 0.2004,
      "step": 2955
    },
    {
      "epoch": 3.74,
      "learning_rate": 2.3273056057866186e-05,
      "loss": 0.0025,
      "step": 2956
    },
    {
      "epoch": 3.74,
      "learning_rate": 2.3264014466546114e-05,
      "loss": 0.0604,
      "step": 2957
    },
    {
      "epoch": 3.74,
      "learning_rate": 2.3254972875226043e-05,
      "loss": 0.117,
      "step": 2958
    },
    {
      "epoch": 3.75,
      "learning_rate": 2.324593128390597e-05,
      "loss": 0.011,
      "step": 2959
    },
    {
      "epoch": 3.75,
      "learning_rate": 2.32368896925859e-05,
      "loss": 0.026,
      "step": 2960
    },
    {
      "epoch": 3.75,
      "learning_rate": 2.3227848101265824e-05,
      "loss": 0.0439,
      "step": 2961
    },
    {
      "epoch": 3.75,
      "learning_rate": 2.3218806509945752e-05,
      "loss": 0.3097,
      "step": 2962
    },
    {
      "epoch": 3.75,
      "learning_rate": 2.320976491862568e-05,
      "loss": 0.0068,
      "step": 2963
    },
    {
      "epoch": 3.75,
      "learning_rate": 2.3200723327305605e-05,
      "loss": 0.003,
      "step": 2964
    },
    {
      "epoch": 3.75,
      "learning_rate": 2.3191681735985534e-05,
      "loss": 0.0732,
      "step": 2965
    },
    {
      "epoch": 3.75,
      "learning_rate": 2.3182640144665462e-05,
      "loss": 0.1882,
      "step": 2966
    },
    {
      "epoch": 3.76,
      "learning_rate": 2.317359855334539e-05,
      "loss": 0.0011,
      "step": 2967
    },
    {
      "epoch": 3.76,
      "learning_rate": 2.3164556962025318e-05,
      "loss": 0.0021,
      "step": 2968
    },
    {
      "epoch": 3.76,
      "learning_rate": 2.3155515370705243e-05,
      "loss": 0.0041,
      "step": 2969
    },
    {
      "epoch": 3.76,
      "learning_rate": 2.314647377938517e-05,
      "loss": 0.0089,
      "step": 2970
    },
    {
      "epoch": 3.76,
      "learning_rate": 2.31374321880651e-05,
      "loss": 0.011,
      "step": 2971
    },
    {
      "epoch": 3.76,
      "learning_rate": 2.3128390596745028e-05,
      "loss": 0.042,
      "step": 2972
    },
    {
      "epoch": 3.76,
      "learning_rate": 2.3119349005424956e-05,
      "loss": 0.0042,
      "step": 2973
    },
    {
      "epoch": 3.76,
      "learning_rate": 2.3110307414104884e-05,
      "loss": 0.0018,
      "step": 2974
    },
    {
      "epoch": 3.77,
      "learning_rate": 2.3101265822784813e-05,
      "loss": 0.0957,
      "step": 2975
    },
    {
      "epoch": 3.77,
      "learning_rate": 2.309222423146474e-05,
      "loss": 0.0157,
      "step": 2976
    },
    {
      "epoch": 3.77,
      "learning_rate": 2.3083182640144666e-05,
      "loss": 0.0098,
      "step": 2977
    },
    {
      "epoch": 3.77,
      "learning_rate": 2.3074141048824594e-05,
      "loss": 0.004,
      "step": 2978
    },
    {
      "epoch": 3.77,
      "learning_rate": 2.3065099457504522e-05,
      "loss": 0.0413,
      "step": 2979
    },
    {
      "epoch": 3.77,
      "learning_rate": 2.305605786618445e-05,
      "loss": 0.0799,
      "step": 2980
    },
    {
      "epoch": 3.77,
      "learning_rate": 2.304701627486438e-05,
      "loss": 0.1879,
      "step": 2981
    },
    {
      "epoch": 3.77,
      "learning_rate": 2.3037974683544307e-05,
      "loss": 0.21,
      "step": 2982
    },
    {
      "epoch": 3.78,
      "learning_rate": 2.3028933092224235e-05,
      "loss": 0.069,
      "step": 2983
    },
    {
      "epoch": 3.78,
      "learning_rate": 2.301989150090416e-05,
      "loss": 0.0906,
      "step": 2984
    },
    {
      "epoch": 3.78,
      "learning_rate": 2.3010849909584088e-05,
      "loss": 0.0449,
      "step": 2985
    },
    {
      "epoch": 3.78,
      "learning_rate": 2.3001808318264016e-05,
      "loss": 0.0411,
      "step": 2986
    },
    {
      "epoch": 3.78,
      "learning_rate": 2.299276672694394e-05,
      "loss": 0.0043,
      "step": 2987
    },
    {
      "epoch": 3.78,
      "learning_rate": 2.298372513562387e-05,
      "loss": 0.0306,
      "step": 2988
    },
    {
      "epoch": 3.78,
      "learning_rate": 2.2974683544303798e-05,
      "loss": 0.0686,
      "step": 2989
    },
    {
      "epoch": 3.78,
      "learning_rate": 2.2965641952983726e-05,
      "loss": 0.0014,
      "step": 2990
    },
    {
      "epoch": 3.79,
      "learning_rate": 2.2956600361663654e-05,
      "loss": 0.0058,
      "step": 2991
    },
    {
      "epoch": 3.79,
      "learning_rate": 2.2947558770343582e-05,
      "loss": 0.0084,
      "step": 2992
    },
    {
      "epoch": 3.79,
      "learning_rate": 2.2938517179023507e-05,
      "loss": 0.0112,
      "step": 2993
    },
    {
      "epoch": 3.79,
      "learning_rate": 2.2929475587703436e-05,
      "loss": 0.0287,
      "step": 2994
    },
    {
      "epoch": 3.79,
      "learning_rate": 2.2920433996383364e-05,
      "loss": 0.0098,
      "step": 2995
    },
    {
      "epoch": 3.79,
      "learning_rate": 2.2911392405063292e-05,
      "loss": 0.0514,
      "step": 2996
    },
    {
      "epoch": 3.79,
      "learning_rate": 2.290235081374322e-05,
      "loss": 0.0008,
      "step": 2997
    },
    {
      "epoch": 3.79,
      "learning_rate": 2.289330922242315e-05,
      "loss": 0.037,
      "step": 2998
    },
    {
      "epoch": 3.8,
      "learning_rate": 2.2884267631103077e-05,
      "loss": 0.0956,
      "step": 2999
    },
    {
      "epoch": 3.8,
      "learning_rate": 2.2875226039783005e-05,
      "loss": 0.0862,
      "step": 3000
    },
    {
      "epoch": 3.8,
      "eval_accuracy": 0.9114114114114115,
      "eval_loss": 0.40487927198410034,
      "eval_runtime": 45.0875,
      "eval_samples_per_second": 29.543,
      "eval_steps_per_second": 3.704,
      "step": 3000
    },
    {
      "epoch": 3.8,
      "learning_rate": 2.286618444846293e-05,
      "loss": 0.0041,
      "step": 3001
    },
    {
      "epoch": 3.8,
      "learning_rate": 2.2857142857142858e-05,
      "loss": 0.0943,
      "step": 3002
    },
    {
      "epoch": 3.8,
      "learning_rate": 2.2848101265822786e-05,
      "loss": 0.0009,
      "step": 3003
    },
    {
      "epoch": 3.8,
      "learning_rate": 2.2839059674502715e-05,
      "loss": 0.0024,
      "step": 3004
    },
    {
      "epoch": 3.8,
      "learning_rate": 2.2830018083182643e-05,
      "loss": 0.0161,
      "step": 3005
    },
    {
      "epoch": 3.8,
      "learning_rate": 2.282097649186257e-05,
      "loss": 0.0572,
      "step": 3006
    },
    {
      "epoch": 3.81,
      "learning_rate": 2.28119349005425e-05,
      "loss": 0.1528,
      "step": 3007
    },
    {
      "epoch": 3.81,
      "learning_rate": 2.2802893309222424e-05,
      "loss": 0.0867,
      "step": 3008
    },
    {
      "epoch": 3.81,
      "learning_rate": 2.2793851717902352e-05,
      "loss": 0.1089,
      "step": 3009
    },
    {
      "epoch": 3.81,
      "learning_rate": 2.278481012658228e-05,
      "loss": 0.0019,
      "step": 3010
    },
    {
      "epoch": 3.81,
      "learning_rate": 2.2775768535262205e-05,
      "loss": 0.0803,
      "step": 3011
    },
    {
      "epoch": 3.81,
      "learning_rate": 2.2766726943942134e-05,
      "loss": 0.0132,
      "step": 3012
    },
    {
      "epoch": 3.81,
      "learning_rate": 2.2757685352622062e-05,
      "loss": 0.0041,
      "step": 3013
    },
    {
      "epoch": 3.81,
      "learning_rate": 2.274864376130199e-05,
      "loss": 0.0018,
      "step": 3014
    },
    {
      "epoch": 3.82,
      "learning_rate": 2.273960216998192e-05,
      "loss": 0.0423,
      "step": 3015
    },
    {
      "epoch": 3.82,
      "learning_rate": 2.2730560578661847e-05,
      "loss": 0.0023,
      "step": 3016
    },
    {
      "epoch": 3.82,
      "learning_rate": 2.272151898734177e-05,
      "loss": 0.0563,
      "step": 3017
    },
    {
      "epoch": 3.82,
      "learning_rate": 2.27124773960217e-05,
      "loss": 0.0095,
      "step": 3018
    },
    {
      "epoch": 3.82,
      "learning_rate": 2.2703435804701628e-05,
      "loss": 0.0179,
      "step": 3019
    },
    {
      "epoch": 3.82,
      "learning_rate": 2.2694394213381556e-05,
      "loss": 0.0074,
      "step": 3020
    },
    {
      "epoch": 3.82,
      "learning_rate": 2.2685352622061484e-05,
      "loss": 0.0399,
      "step": 3021
    },
    {
      "epoch": 3.82,
      "learning_rate": 2.2676311030741413e-05,
      "loss": 0.0018,
      "step": 3022
    },
    {
      "epoch": 3.83,
      "learning_rate": 2.266726943942134e-05,
      "loss": 0.0145,
      "step": 3023
    },
    {
      "epoch": 3.83,
      "learning_rate": 2.2658227848101266e-05,
      "loss": 0.0041,
      "step": 3024
    },
    {
      "epoch": 3.83,
      "learning_rate": 2.2649186256781194e-05,
      "loss": 0.0183,
      "step": 3025
    },
    {
      "epoch": 3.83,
      "learning_rate": 2.2640144665461122e-05,
      "loss": 0.0025,
      "step": 3026
    },
    {
      "epoch": 3.83,
      "learning_rate": 2.263110307414105e-05,
      "loss": 0.1715,
      "step": 3027
    },
    {
      "epoch": 3.83,
      "learning_rate": 2.262206148282098e-05,
      "loss": 0.0006,
      "step": 3028
    },
    {
      "epoch": 3.83,
      "learning_rate": 2.2613019891500907e-05,
      "loss": 0.1748,
      "step": 3029
    },
    {
      "epoch": 3.83,
      "learning_rate": 2.2603978300180835e-05,
      "loss": 0.0006,
      "step": 3030
    },
    {
      "epoch": 3.84,
      "learning_rate": 2.259493670886076e-05,
      "loss": 0.0045,
      "step": 3031
    },
    {
      "epoch": 3.84,
      "learning_rate": 2.2585895117540688e-05,
      "loss": 0.0927,
      "step": 3032
    },
    {
      "epoch": 3.84,
      "learning_rate": 2.2576853526220616e-05,
      "loss": 0.1846,
      "step": 3033
    },
    {
      "epoch": 3.84,
      "learning_rate": 2.256781193490054e-05,
      "loss": 0.1374,
      "step": 3034
    },
    {
      "epoch": 3.84,
      "learning_rate": 2.255877034358047e-05,
      "loss": 0.0297,
      "step": 3035
    },
    {
      "epoch": 3.84,
      "learning_rate": 2.2549728752260398e-05,
      "loss": 0.0043,
      "step": 3036
    },
    {
      "epoch": 3.84,
      "learning_rate": 2.2540687160940326e-05,
      "loss": 0.039,
      "step": 3037
    },
    {
      "epoch": 3.85,
      "learning_rate": 2.2531645569620254e-05,
      "loss": 0.0085,
      "step": 3038
    },
    {
      "epoch": 3.85,
      "learning_rate": 2.2522603978300183e-05,
      "loss": 0.0168,
      "step": 3039
    },
    {
      "epoch": 3.85,
      "learning_rate": 2.2513562386980107e-05,
      "loss": 0.1439,
      "step": 3040
    },
    {
      "epoch": 3.85,
      "learning_rate": 2.2504520795660036e-05,
      "loss": 0.031,
      "step": 3041
    },
    {
      "epoch": 3.85,
      "learning_rate": 2.2495479204339964e-05,
      "loss": 0.0379,
      "step": 3042
    },
    {
      "epoch": 3.85,
      "learning_rate": 2.2486437613019892e-05,
      "loss": 0.0222,
      "step": 3043
    },
    {
      "epoch": 3.85,
      "learning_rate": 2.247739602169982e-05,
      "loss": 0.0177,
      "step": 3044
    },
    {
      "epoch": 3.85,
      "learning_rate": 2.246835443037975e-05,
      "loss": 0.0055,
      "step": 3045
    },
    {
      "epoch": 3.86,
      "learning_rate": 2.2459312839059677e-05,
      "loss": 0.0743,
      "step": 3046
    },
    {
      "epoch": 3.86,
      "learning_rate": 2.2450271247739605e-05,
      "loss": 0.08,
      "step": 3047
    },
    {
      "epoch": 3.86,
      "learning_rate": 2.244122965641953e-05,
      "loss": 0.3317,
      "step": 3048
    },
    {
      "epoch": 3.86,
      "learning_rate": 2.2432188065099458e-05,
      "loss": 0.0023,
      "step": 3049
    },
    {
      "epoch": 3.86,
      "learning_rate": 2.2423146473779386e-05,
      "loss": 0.0076,
      "step": 3050
    },
    {
      "epoch": 3.86,
      "learning_rate": 2.2414104882459315e-05,
      "loss": 0.0016,
      "step": 3051
    },
    {
      "epoch": 3.86,
      "learning_rate": 2.2405063291139243e-05,
      "loss": 0.0032,
      "step": 3052
    },
    {
      "epoch": 3.86,
      "learning_rate": 2.239602169981917e-05,
      "loss": 0.2384,
      "step": 3053
    },
    {
      "epoch": 3.87,
      "learning_rate": 2.23869801084991e-05,
      "loss": 0.0178,
      "step": 3054
    },
    {
      "epoch": 3.87,
      "learning_rate": 2.2377938517179024e-05,
      "loss": 0.0154,
      "step": 3055
    },
    {
      "epoch": 3.87,
      "learning_rate": 2.2368896925858952e-05,
      "loss": 0.0812,
      "step": 3056
    },
    {
      "epoch": 3.87,
      "learning_rate": 2.2359855334538877e-05,
      "loss": 0.2117,
      "step": 3057
    },
    {
      "epoch": 3.87,
      "learning_rate": 2.2350813743218805e-05,
      "loss": 0.0191,
      "step": 3058
    },
    {
      "epoch": 3.87,
      "learning_rate": 2.2341772151898734e-05,
      "loss": 0.038,
      "step": 3059
    },
    {
      "epoch": 3.87,
      "learning_rate": 2.2332730560578662e-05,
      "loss": 0.0208,
      "step": 3060
    },
    {
      "epoch": 3.87,
      "learning_rate": 2.232368896925859e-05,
      "loss": 0.2016,
      "step": 3061
    },
    {
      "epoch": 3.88,
      "learning_rate": 2.231464737793852e-05,
      "loss": 0.0093,
      "step": 3062
    },
    {
      "epoch": 3.88,
      "learning_rate": 2.2305605786618447e-05,
      "loss": 0.0027,
      "step": 3063
    },
    {
      "epoch": 3.88,
      "learning_rate": 2.229656419529837e-05,
      "loss": 0.0103,
      "step": 3064
    },
    {
      "epoch": 3.88,
      "learning_rate": 2.22875226039783e-05,
      "loss": 0.0046,
      "step": 3065
    },
    {
      "epoch": 3.88,
      "learning_rate": 2.2278481012658228e-05,
      "loss": 0.1397,
      "step": 3066
    },
    {
      "epoch": 3.88,
      "learning_rate": 2.2269439421338156e-05,
      "loss": 0.0029,
      "step": 3067
    },
    {
      "epoch": 3.88,
      "learning_rate": 2.2260397830018084e-05,
      "loss": 0.1138,
      "step": 3068
    },
    {
      "epoch": 3.88,
      "learning_rate": 2.2251356238698013e-05,
      "loss": 0.0313,
      "step": 3069
    },
    {
      "epoch": 3.89,
      "learning_rate": 2.224231464737794e-05,
      "loss": 0.0013,
      "step": 3070
    },
    {
      "epoch": 3.89,
      "learning_rate": 2.223327305605787e-05,
      "loss": 0.0292,
      "step": 3071
    },
    {
      "epoch": 3.89,
      "learning_rate": 2.2224231464737794e-05,
      "loss": 0.0488,
      "step": 3072
    },
    {
      "epoch": 3.89,
      "learning_rate": 2.2215189873417722e-05,
      "loss": 0.0428,
      "step": 3073
    },
    {
      "epoch": 3.89,
      "learning_rate": 2.220614828209765e-05,
      "loss": 0.0286,
      "step": 3074
    },
    {
      "epoch": 3.89,
      "learning_rate": 2.219710669077758e-05,
      "loss": 0.0315,
      "step": 3075
    },
    {
      "epoch": 3.89,
      "learning_rate": 2.2188065099457507e-05,
      "loss": 0.0022,
      "step": 3076
    },
    {
      "epoch": 3.89,
      "learning_rate": 2.2179023508137435e-05,
      "loss": 0.001,
      "step": 3077
    },
    {
      "epoch": 3.9,
      "learning_rate": 2.216998191681736e-05,
      "loss": 0.0815,
      "step": 3078
    },
    {
      "epoch": 3.9,
      "learning_rate": 2.216094032549729e-05,
      "loss": 0.1473,
      "step": 3079
    },
    {
      "epoch": 3.9,
      "learning_rate": 2.2151898734177217e-05,
      "loss": 0.0041,
      "step": 3080
    },
    {
      "epoch": 3.9,
      "learning_rate": 2.214285714285714e-05,
      "loss": 0.0673,
      "step": 3081
    },
    {
      "epoch": 3.9,
      "learning_rate": 2.213381555153707e-05,
      "loss": 0.0175,
      "step": 3082
    },
    {
      "epoch": 3.9,
      "learning_rate": 2.2124773960216998e-05,
      "loss": 0.0028,
      "step": 3083
    },
    {
      "epoch": 3.9,
      "learning_rate": 2.2115732368896926e-05,
      "loss": 0.0863,
      "step": 3084
    },
    {
      "epoch": 3.9,
      "learning_rate": 2.2106690777576854e-05,
      "loss": 0.1799,
      "step": 3085
    },
    {
      "epoch": 3.91,
      "learning_rate": 2.2097649186256783e-05,
      "loss": 0.0146,
      "step": 3086
    },
    {
      "epoch": 3.91,
      "learning_rate": 2.208860759493671e-05,
      "loss": 0.0434,
      "step": 3087
    },
    {
      "epoch": 3.91,
      "learning_rate": 2.2079566003616636e-05,
      "loss": 0.1412,
      "step": 3088
    },
    {
      "epoch": 3.91,
      "learning_rate": 2.2070524412296564e-05,
      "loss": 0.0135,
      "step": 3089
    },
    {
      "epoch": 3.91,
      "learning_rate": 2.2061482820976492e-05,
      "loss": 0.0023,
      "step": 3090
    },
    {
      "epoch": 3.91,
      "learning_rate": 2.205244122965642e-05,
      "loss": 0.0703,
      "step": 3091
    },
    {
      "epoch": 3.91,
      "learning_rate": 2.204339963833635e-05,
      "loss": 0.1076,
      "step": 3092
    },
    {
      "epoch": 3.91,
      "learning_rate": 2.2034358047016277e-05,
      "loss": 0.0138,
      "step": 3093
    },
    {
      "epoch": 3.92,
      "learning_rate": 2.2025316455696205e-05,
      "loss": 0.0049,
      "step": 3094
    },
    {
      "epoch": 3.92,
      "learning_rate": 2.2016274864376133e-05,
      "loss": 0.0017,
      "step": 3095
    },
    {
      "epoch": 3.92,
      "learning_rate": 2.2007233273056058e-05,
      "loss": 0.0402,
      "step": 3096
    },
    {
      "epoch": 3.92,
      "learning_rate": 2.1998191681735986e-05,
      "loss": 0.0833,
      "step": 3097
    },
    {
      "epoch": 3.92,
      "learning_rate": 2.1989150090415915e-05,
      "loss": 0.1308,
      "step": 3098
    },
    {
      "epoch": 3.92,
      "learning_rate": 2.1980108499095843e-05,
      "loss": 0.0513,
      "step": 3099
    },
    {
      "epoch": 3.92,
      "learning_rate": 2.197106690777577e-05,
      "loss": 0.0014,
      "step": 3100
    },
    {
      "epoch": 3.92,
      "learning_rate": 2.19620253164557e-05,
      "loss": 0.0261,
      "step": 3101
    },
    {
      "epoch": 3.93,
      "learning_rate": 2.1952983725135624e-05,
      "loss": 0.0819,
      "step": 3102
    },
    {
      "epoch": 3.93,
      "learning_rate": 2.1943942133815552e-05,
      "loss": 0.0283,
      "step": 3103
    },
    {
      "epoch": 3.93,
      "learning_rate": 2.1934900542495477e-05,
      "loss": 0.0052,
      "step": 3104
    },
    {
      "epoch": 3.93,
      "learning_rate": 2.1925858951175406e-05,
      "loss": 0.1646,
      "step": 3105
    },
    {
      "epoch": 3.93,
      "learning_rate": 2.1916817359855334e-05,
      "loss": 0.0565,
      "step": 3106
    },
    {
      "epoch": 3.93,
      "learning_rate": 2.1907775768535262e-05,
      "loss": 0.1656,
      "step": 3107
    },
    {
      "epoch": 3.93,
      "learning_rate": 2.189873417721519e-05,
      "loss": 0.0076,
      "step": 3108
    },
    {
      "epoch": 3.93,
      "learning_rate": 2.188969258589512e-05,
      "loss": 0.0296,
      "step": 3109
    },
    {
      "epoch": 3.94,
      "learning_rate": 2.1880650994575047e-05,
      "loss": 0.1948,
      "step": 3110
    },
    {
      "epoch": 3.94,
      "learning_rate": 2.1871609403254975e-05,
      "loss": 0.0074,
      "step": 3111
    },
    {
      "epoch": 3.94,
      "learning_rate": 2.18625678119349e-05,
      "loss": 0.1695,
      "step": 3112
    },
    {
      "epoch": 3.94,
      "learning_rate": 2.1853526220614828e-05,
      "loss": 0.0037,
      "step": 3113
    },
    {
      "epoch": 3.94,
      "learning_rate": 2.1844484629294756e-05,
      "loss": 0.1644,
      "step": 3114
    },
    {
      "epoch": 3.94,
      "learning_rate": 2.1835443037974685e-05,
      "loss": 0.1069,
      "step": 3115
    },
    {
      "epoch": 3.94,
      "learning_rate": 2.1826401446654613e-05,
      "loss": 0.2496,
      "step": 3116
    },
    {
      "epoch": 3.94,
      "learning_rate": 2.181735985533454e-05,
      "loss": 0.2344,
      "step": 3117
    },
    {
      "epoch": 3.95,
      "learning_rate": 2.180831826401447e-05,
      "loss": 0.0617,
      "step": 3118
    },
    {
      "epoch": 3.95,
      "learning_rate": 2.1799276672694397e-05,
      "loss": 0.0047,
      "step": 3119
    },
    {
      "epoch": 3.95,
      "learning_rate": 2.1790235081374322e-05,
      "loss": 0.1041,
      "step": 3120
    },
    {
      "epoch": 3.95,
      "learning_rate": 2.178119349005425e-05,
      "loss": 0.0233,
      "step": 3121
    },
    {
      "epoch": 3.95,
      "learning_rate": 2.177215189873418e-05,
      "loss": 0.0085,
      "step": 3122
    },
    {
      "epoch": 3.95,
      "learning_rate": 2.1763110307414107e-05,
      "loss": 0.0048,
      "step": 3123
    },
    {
      "epoch": 3.95,
      "learning_rate": 2.1754068716094035e-05,
      "loss": 0.0225,
      "step": 3124
    },
    {
      "epoch": 3.96,
      "learning_rate": 2.174502712477396e-05,
      "loss": 0.3527,
      "step": 3125
    },
    {
      "epoch": 3.96,
      "learning_rate": 2.173598553345389e-05,
      "loss": 0.0018,
      "step": 3126
    },
    {
      "epoch": 3.96,
      "learning_rate": 2.1726943942133817e-05,
      "loss": 0.0365,
      "step": 3127
    },
    {
      "epoch": 3.96,
      "learning_rate": 2.171790235081374e-05,
      "loss": 0.0034,
      "step": 3128
    },
    {
      "epoch": 3.96,
      "learning_rate": 2.170886075949367e-05,
      "loss": 0.0111,
      "step": 3129
    },
    {
      "epoch": 3.96,
      "learning_rate": 2.1699819168173598e-05,
      "loss": 0.097,
      "step": 3130
    },
    {
      "epoch": 3.96,
      "learning_rate": 2.1690777576853526e-05,
      "loss": 0.2383,
      "step": 3131
    },
    {
      "epoch": 3.96,
      "learning_rate": 2.1681735985533454e-05,
      "loss": 0.0025,
      "step": 3132
    },
    {
      "epoch": 3.97,
      "learning_rate": 2.1672694394213383e-05,
      "loss": 0.1137,
      "step": 3133
    },
    {
      "epoch": 3.97,
      "learning_rate": 2.166365280289331e-05,
      "loss": 0.1051,
      "step": 3134
    },
    {
      "epoch": 3.97,
      "learning_rate": 2.165461121157324e-05,
      "loss": 0.1682,
      "step": 3135
    },
    {
      "epoch": 3.97,
      "learning_rate": 2.1645569620253164e-05,
      "loss": 0.0094,
      "step": 3136
    },
    {
      "epoch": 3.97,
      "learning_rate": 2.1636528028933092e-05,
      "loss": 0.0152,
      "step": 3137
    },
    {
      "epoch": 3.97,
      "learning_rate": 2.162748643761302e-05,
      "loss": 0.0048,
      "step": 3138
    },
    {
      "epoch": 3.97,
      "learning_rate": 2.161844484629295e-05,
      "loss": 0.0059,
      "step": 3139
    },
    {
      "epoch": 3.97,
      "learning_rate": 2.1609403254972877e-05,
      "loss": 0.0252,
      "step": 3140
    },
    {
      "epoch": 3.98,
      "learning_rate": 2.1600361663652805e-05,
      "loss": 0.0033,
      "step": 3141
    },
    {
      "epoch": 3.98,
      "learning_rate": 2.1591320072332733e-05,
      "loss": 0.0579,
      "step": 3142
    },
    {
      "epoch": 3.98,
      "learning_rate": 2.158227848101266e-05,
      "loss": 0.0048,
      "step": 3143
    },
    {
      "epoch": 3.98,
      "learning_rate": 2.1573236889692587e-05,
      "loss": 0.0037,
      "step": 3144
    },
    {
      "epoch": 3.98,
      "learning_rate": 2.1564195298372515e-05,
      "loss": 0.3841,
      "step": 3145
    },
    {
      "epoch": 3.98,
      "learning_rate": 2.1555153707052443e-05,
      "loss": 0.1822,
      "step": 3146
    },
    {
      "epoch": 3.98,
      "learning_rate": 2.154611211573237e-05,
      "loss": 0.2192,
      "step": 3147
    },
    {
      "epoch": 3.98,
      "learning_rate": 2.15370705244123e-05,
      "loss": 0.0681,
      "step": 3148
    },
    {
      "epoch": 3.99,
      "learning_rate": 2.1528028933092224e-05,
      "loss": 0.0413,
      "step": 3149
    },
    {
      "epoch": 3.99,
      "learning_rate": 2.1518987341772153e-05,
      "loss": 0.0042,
      "step": 3150
    },
    {
      "epoch": 3.99,
      "learning_rate": 2.150994575045208e-05,
      "loss": 0.1854,
      "step": 3151
    },
    {
      "epoch": 3.99,
      "learning_rate": 2.1500904159132006e-05,
      "loss": 0.0115,
      "step": 3152
    },
    {
      "epoch": 3.99,
      "learning_rate": 2.1491862567811934e-05,
      "loss": 0.0049,
      "step": 3153
    },
    {
      "epoch": 3.99,
      "learning_rate": 2.1482820976491862e-05,
      "loss": 0.0324,
      "step": 3154
    },
    {
      "epoch": 3.99,
      "learning_rate": 2.147377938517179e-05,
      "loss": 0.0065,
      "step": 3155
    },
    {
      "epoch": 3.99,
      "learning_rate": 2.146473779385172e-05,
      "loss": 0.0367,
      "step": 3156
    },
    {
      "epoch": 4.0,
      "learning_rate": 2.1455696202531647e-05,
      "loss": 0.1623,
      "step": 3157
    },
    {
      "epoch": 4.0,
      "learning_rate": 2.1446654611211575e-05,
      "loss": 0.0047,
      "step": 3158
    },
    {
      "epoch": 4.0,
      "learning_rate": 2.1437613019891503e-05,
      "loss": 0.0083,
      "step": 3159
    },
    {
      "epoch": 4.0,
      "learning_rate": 2.1428571428571428e-05,
      "loss": 0.1612,
      "step": 3160
    },
    {
      "epoch": 4.0,
      "learning_rate": 2.1419529837251356e-05,
      "loss": 0.0076,
      "step": 3161
    },
    {
      "epoch": 4.0,
      "learning_rate": 2.1410488245931285e-05,
      "loss": 0.0057,
      "step": 3162
    },
    {
      "epoch": 4.0,
      "learning_rate": 2.1401446654611213e-05,
      "loss": 0.0542,
      "step": 3163
    },
    {
      "epoch": 4.01,
      "learning_rate": 2.139240506329114e-05,
      "loss": 0.0024,
      "step": 3164
    },
    {
      "epoch": 4.01,
      "learning_rate": 2.138336347197107e-05,
      "loss": 0.0013,
      "step": 3165
    },
    {
      "epoch": 4.01,
      "learning_rate": 2.1374321880650998e-05,
      "loss": 0.1158,
      "step": 3166
    },
    {
      "epoch": 4.01,
      "learning_rate": 2.1365280289330926e-05,
      "loss": 0.0207,
      "step": 3167
    },
    {
      "epoch": 4.01,
      "learning_rate": 2.135623869801085e-05,
      "loss": 0.0054,
      "step": 3168
    },
    {
      "epoch": 4.01,
      "learning_rate": 2.134719710669078e-05,
      "loss": 0.0098,
      "step": 3169
    },
    {
      "epoch": 4.01,
      "learning_rate": 2.1338155515370707e-05,
      "loss": 0.0015,
      "step": 3170
    },
    {
      "epoch": 4.01,
      "learning_rate": 2.1329113924050635e-05,
      "loss": 0.0636,
      "step": 3171
    },
    {
      "epoch": 4.02,
      "learning_rate": 2.132007233273056e-05,
      "loss": 0.0403,
      "step": 3172
    },
    {
      "epoch": 4.02,
      "learning_rate": 2.131103074141049e-05,
      "loss": 0.0041,
      "step": 3173
    },
    {
      "epoch": 4.02,
      "learning_rate": 2.1301989150090417e-05,
      "loss": 0.003,
      "step": 3174
    },
    {
      "epoch": 4.02,
      "learning_rate": 2.1292947558770345e-05,
      "loss": 0.0125,
      "step": 3175
    },
    {
      "epoch": 4.02,
      "learning_rate": 2.128390596745027e-05,
      "loss": 0.0074,
      "step": 3176
    },
    {
      "epoch": 4.02,
      "learning_rate": 2.1274864376130198e-05,
      "loss": 0.0011,
      "step": 3177
    },
    {
      "epoch": 4.02,
      "learning_rate": 2.1265822784810126e-05,
      "loss": 0.0063,
      "step": 3178
    },
    {
      "epoch": 4.02,
      "learning_rate": 2.1256781193490055e-05,
      "loss": 0.002,
      "step": 3179
    },
    {
      "epoch": 4.03,
      "learning_rate": 2.1247739602169983e-05,
      "loss": 0.0043,
      "step": 3180
    },
    {
      "epoch": 4.03,
      "learning_rate": 2.123869801084991e-05,
      "loss": 0.002,
      "step": 3181
    },
    {
      "epoch": 4.03,
      "learning_rate": 2.122965641952984e-05,
      "loss": 0.0015,
      "step": 3182
    },
    {
      "epoch": 4.03,
      "learning_rate": 2.1220614828209767e-05,
      "loss": 0.0031,
      "step": 3183
    },
    {
      "epoch": 4.03,
      "learning_rate": 2.1211573236889692e-05,
      "loss": 0.0023,
      "step": 3184
    },
    {
      "epoch": 4.03,
      "learning_rate": 2.120253164556962e-05,
      "loss": 0.2164,
      "step": 3185
    },
    {
      "epoch": 4.03,
      "learning_rate": 2.119349005424955e-05,
      "loss": 0.0045,
      "step": 3186
    },
    {
      "epoch": 4.03,
      "learning_rate": 2.1184448462929477e-05,
      "loss": 0.0166,
      "step": 3187
    },
    {
      "epoch": 4.04,
      "learning_rate": 2.1175406871609405e-05,
      "loss": 0.002,
      "step": 3188
    },
    {
      "epoch": 4.04,
      "learning_rate": 2.1166365280289333e-05,
      "loss": 0.0016,
      "step": 3189
    },
    {
      "epoch": 4.04,
      "learning_rate": 2.1157323688969262e-05,
      "loss": 0.1056,
      "step": 3190
    },
    {
      "epoch": 4.04,
      "learning_rate": 2.114828209764919e-05,
      "loss": 0.0358,
      "step": 3191
    },
    {
      "epoch": 4.04,
      "learning_rate": 2.1139240506329115e-05,
      "loss": 0.0302,
      "step": 3192
    },
    {
      "epoch": 4.04,
      "learning_rate": 2.1130198915009043e-05,
      "loss": 0.0024,
      "step": 3193
    },
    {
      "epoch": 4.04,
      "learning_rate": 2.112115732368897e-05,
      "loss": 0.0019,
      "step": 3194
    },
    {
      "epoch": 4.04,
      "learning_rate": 2.11121157323689e-05,
      "loss": 0.019,
      "step": 3195
    },
    {
      "epoch": 4.05,
      "learning_rate": 2.1103074141048824e-05,
      "loss": 0.0062,
      "step": 3196
    },
    {
      "epoch": 4.05,
      "learning_rate": 2.1094032549728753e-05,
      "loss": 0.003,
      "step": 3197
    },
    {
      "epoch": 4.05,
      "learning_rate": 2.108499095840868e-05,
      "loss": 0.1142,
      "step": 3198
    },
    {
      "epoch": 4.05,
      "learning_rate": 2.107594936708861e-05,
      "loss": 0.0431,
      "step": 3199
    },
    {
      "epoch": 4.05,
      "learning_rate": 2.1066907775768534e-05,
      "loss": 0.0034,
      "step": 3200
    },
    {
      "epoch": 4.05,
      "learning_rate": 2.1057866184448462e-05,
      "loss": 0.0592,
      "step": 3201
    },
    {
      "epoch": 4.05,
      "learning_rate": 2.104882459312839e-05,
      "loss": 0.0013,
      "step": 3202
    },
    {
      "epoch": 4.05,
      "learning_rate": 2.103978300180832e-05,
      "loss": 0.0018,
      "step": 3203
    },
    {
      "epoch": 4.06,
      "learning_rate": 2.1030741410488247e-05,
      "loss": 0.0016,
      "step": 3204
    },
    {
      "epoch": 4.06,
      "learning_rate": 2.1021699819168175e-05,
      "loss": 0.0067,
      "step": 3205
    },
    {
      "epoch": 4.06,
      "learning_rate": 2.1012658227848103e-05,
      "loss": 0.0054,
      "step": 3206
    },
    {
      "epoch": 4.06,
      "learning_rate": 2.100361663652803e-05,
      "loss": 0.0179,
      "step": 3207
    },
    {
      "epoch": 4.06,
      "learning_rate": 2.0994575045207956e-05,
      "loss": 0.0008,
      "step": 3208
    },
    {
      "epoch": 4.06,
      "learning_rate": 2.0985533453887885e-05,
      "loss": 0.0012,
      "step": 3209
    },
    {
      "epoch": 4.06,
      "learning_rate": 2.0976491862567813e-05,
      "loss": 0.0121,
      "step": 3210
    },
    {
      "epoch": 4.06,
      "learning_rate": 2.096745027124774e-05,
      "loss": 0.0105,
      "step": 3211
    },
    {
      "epoch": 4.07,
      "learning_rate": 2.095840867992767e-05,
      "loss": 0.0639,
      "step": 3212
    },
    {
      "epoch": 4.07,
      "learning_rate": 2.0949367088607598e-05,
      "loss": 0.0065,
      "step": 3213
    },
    {
      "epoch": 4.07,
      "learning_rate": 2.0940325497287526e-05,
      "loss": 0.0021,
      "step": 3214
    },
    {
      "epoch": 4.07,
      "learning_rate": 2.0931283905967454e-05,
      "loss": 0.0053,
      "step": 3215
    },
    {
      "epoch": 4.07,
      "learning_rate": 2.092224231464738e-05,
      "loss": 0.1873,
      "step": 3216
    },
    {
      "epoch": 4.07,
      "learning_rate": 2.0913200723327307e-05,
      "loss": 0.0012,
      "step": 3217
    },
    {
      "epoch": 4.07,
      "learning_rate": 2.0904159132007235e-05,
      "loss": 0.0044,
      "step": 3218
    },
    {
      "epoch": 4.07,
      "learning_rate": 2.089511754068716e-05,
      "loss": 0.0012,
      "step": 3219
    },
    {
      "epoch": 4.08,
      "learning_rate": 2.088607594936709e-05,
      "loss": 0.0013,
      "step": 3220
    },
    {
      "epoch": 4.08,
      "learning_rate": 2.0877034358047017e-05,
      "loss": 0.0011,
      "step": 3221
    },
    {
      "epoch": 4.08,
      "learning_rate": 2.0867992766726945e-05,
      "loss": 0.0059,
      "step": 3222
    },
    {
      "epoch": 4.08,
      "learning_rate": 2.0858951175406873e-05,
      "loss": 0.0013,
      "step": 3223
    },
    {
      "epoch": 4.08,
      "learning_rate": 2.0849909584086798e-05,
      "loss": 0.0048,
      "step": 3224
    },
    {
      "epoch": 4.08,
      "learning_rate": 2.0840867992766726e-05,
      "loss": 0.0041,
      "step": 3225
    },
    {
      "epoch": 4.08,
      "learning_rate": 2.0831826401446655e-05,
      "loss": 0.0915,
      "step": 3226
    },
    {
      "epoch": 4.08,
      "learning_rate": 2.0822784810126583e-05,
      "loss": 0.025,
      "step": 3227
    },
    {
      "epoch": 4.09,
      "learning_rate": 2.081374321880651e-05,
      "loss": 0.1219,
      "step": 3228
    },
    {
      "epoch": 4.09,
      "learning_rate": 2.080470162748644e-05,
      "loss": 0.0008,
      "step": 3229
    },
    {
      "epoch": 4.09,
      "learning_rate": 2.0795660036166368e-05,
      "loss": 0.0316,
      "step": 3230
    },
    {
      "epoch": 4.09,
      "learning_rate": 2.0786618444846296e-05,
      "loss": 0.0582,
      "step": 3231
    },
    {
      "epoch": 4.09,
      "learning_rate": 2.077757685352622e-05,
      "loss": 0.0036,
      "step": 3232
    },
    {
      "epoch": 4.09,
      "learning_rate": 2.076853526220615e-05,
      "loss": 0.0117,
      "step": 3233
    },
    {
      "epoch": 4.09,
      "learning_rate": 2.0759493670886077e-05,
      "loss": 0.0654,
      "step": 3234
    },
    {
      "epoch": 4.09,
      "learning_rate": 2.0750452079566005e-05,
      "loss": 0.0024,
      "step": 3235
    },
    {
      "epoch": 4.1,
      "learning_rate": 2.0741410488245934e-05,
      "loss": 0.0038,
      "step": 3236
    },
    {
      "epoch": 4.1,
      "learning_rate": 2.0732368896925862e-05,
      "loss": 0.0104,
      "step": 3237
    },
    {
      "epoch": 4.1,
      "learning_rate": 2.072332730560579e-05,
      "loss": 0.0011,
      "step": 3238
    },
    {
      "epoch": 4.1,
      "learning_rate": 2.0714285714285718e-05,
      "loss": 0.0106,
      "step": 3239
    },
    {
      "epoch": 4.1,
      "learning_rate": 2.0705244122965643e-05,
      "loss": 0.177,
      "step": 3240
    },
    {
      "epoch": 4.1,
      "learning_rate": 2.069620253164557e-05,
      "loss": 0.0009,
      "step": 3241
    },
    {
      "epoch": 4.1,
      "learning_rate": 2.0687160940325496e-05,
      "loss": 0.1115,
      "step": 3242
    },
    {
      "epoch": 4.1,
      "learning_rate": 2.0678119349005424e-05,
      "loss": 0.001,
      "step": 3243
    },
    {
      "epoch": 4.11,
      "learning_rate": 2.0669077757685353e-05,
      "loss": 0.002,
      "step": 3244
    },
    {
      "epoch": 4.11,
      "learning_rate": 2.066003616636528e-05,
      "loss": 0.0096,
      "step": 3245
    },
    {
      "epoch": 4.11,
      "learning_rate": 2.065099457504521e-05,
      "loss": 0.0017,
      "step": 3246
    },
    {
      "epoch": 4.11,
      "learning_rate": 2.0641952983725137e-05,
      "loss": 0.0011,
      "step": 3247
    },
    {
      "epoch": 4.11,
      "learning_rate": 2.0632911392405062e-05,
      "loss": 0.0018,
      "step": 3248
    },
    {
      "epoch": 4.11,
      "learning_rate": 2.062386980108499e-05,
      "loss": 0.0019,
      "step": 3249
    },
    {
      "epoch": 4.11,
      "learning_rate": 2.061482820976492e-05,
      "loss": 0.0017,
      "step": 3250
    },
    {
      "epoch": 4.11,
      "eval_accuracy": 0.9054054054054054,
      "eval_loss": 0.3993643820285797,
      "eval_runtime": 45.1377,
      "eval_samples_per_second": 29.51,
      "eval_steps_per_second": 3.7,
      "step": 3250
    },
    {
      "epoch": 4.12,
      "learning_rate": 2.0605786618444847e-05,
      "loss": 0.0016,
      "step": 3251
    },
    {
      "epoch": 4.12,
      "learning_rate": 2.0596745027124775e-05,
      "loss": 0.0032,
      "step": 3252
    },
    {
      "epoch": 4.12,
      "learning_rate": 2.0587703435804703e-05,
      "loss": 0.1483,
      "step": 3253
    },
    {
      "epoch": 4.12,
      "learning_rate": 2.057866184448463e-05,
      "loss": 0.0016,
      "step": 3254
    },
    {
      "epoch": 4.12,
      "learning_rate": 2.056962025316456e-05,
      "loss": 0.0028,
      "step": 3255
    },
    {
      "epoch": 4.12,
      "learning_rate": 2.0560578661844485e-05,
      "loss": 0.0381,
      "step": 3256
    },
    {
      "epoch": 4.12,
      "learning_rate": 2.0551537070524413e-05,
      "loss": 0.0345,
      "step": 3257
    },
    {
      "epoch": 4.12,
      "learning_rate": 2.054249547920434e-05,
      "loss": 0.0055,
      "step": 3258
    },
    {
      "epoch": 4.13,
      "learning_rate": 2.053345388788427e-05,
      "loss": 0.0097,
      "step": 3259
    },
    {
      "epoch": 4.13,
      "learning_rate": 2.0524412296564198e-05,
      "loss": 0.0172,
      "step": 3260
    },
    {
      "epoch": 4.13,
      "learning_rate": 2.0515370705244126e-05,
      "loss": 0.0037,
      "step": 3261
    },
    {
      "epoch": 4.13,
      "learning_rate": 2.0506329113924054e-05,
      "loss": 0.026,
      "step": 3262
    },
    {
      "epoch": 4.13,
      "learning_rate": 2.049728752260398e-05,
      "loss": 0.0071,
      "step": 3263
    },
    {
      "epoch": 4.13,
      "learning_rate": 2.0488245931283907e-05,
      "loss": 0.0009,
      "step": 3264
    },
    {
      "epoch": 4.13,
      "learning_rate": 2.0479204339963836e-05,
      "loss": 0.001,
      "step": 3265
    },
    {
      "epoch": 4.13,
      "learning_rate": 2.047016274864376e-05,
      "loss": 0.0011,
      "step": 3266
    },
    {
      "epoch": 4.14,
      "learning_rate": 2.046112115732369e-05,
      "loss": 0.0013,
      "step": 3267
    },
    {
      "epoch": 4.14,
      "learning_rate": 2.0452079566003617e-05,
      "loss": 0.0046,
      "step": 3268
    },
    {
      "epoch": 4.14,
      "learning_rate": 2.0443037974683545e-05,
      "loss": 0.036,
      "step": 3269
    },
    {
      "epoch": 4.14,
      "learning_rate": 2.0433996383363473e-05,
      "loss": 0.0011,
      "step": 3270
    },
    {
      "epoch": 4.14,
      "learning_rate": 2.0424954792043398e-05,
      "loss": 0.2547,
      "step": 3271
    },
    {
      "epoch": 4.14,
      "learning_rate": 2.0415913200723326e-05,
      "loss": 0.0017,
      "step": 3272
    },
    {
      "epoch": 4.14,
      "learning_rate": 2.0406871609403255e-05,
      "loss": 0.003,
      "step": 3273
    },
    {
      "epoch": 4.14,
      "learning_rate": 2.0397830018083183e-05,
      "loss": 0.0031,
      "step": 3274
    },
    {
      "epoch": 4.15,
      "learning_rate": 2.038878842676311e-05,
      "loss": 0.002,
      "step": 3275
    },
    {
      "epoch": 4.15,
      "learning_rate": 2.037974683544304e-05,
      "loss": 0.0046,
      "step": 3276
    },
    {
      "epoch": 4.15,
      "learning_rate": 2.0370705244122968e-05,
      "loss": 0.0486,
      "step": 3277
    },
    {
      "epoch": 4.15,
      "learning_rate": 2.0361663652802896e-05,
      "loss": 0.0746,
      "step": 3278
    },
    {
      "epoch": 4.15,
      "learning_rate": 2.035262206148282e-05,
      "loss": 0.0117,
      "step": 3279
    },
    {
      "epoch": 4.15,
      "learning_rate": 2.034358047016275e-05,
      "loss": 0.0014,
      "step": 3280
    },
    {
      "epoch": 4.15,
      "learning_rate": 2.0334538878842677e-05,
      "loss": 0.0022,
      "step": 3281
    },
    {
      "epoch": 4.15,
      "learning_rate": 2.0325497287522605e-05,
      "loss": 0.0117,
      "step": 3282
    },
    {
      "epoch": 4.16,
      "learning_rate": 2.0316455696202534e-05,
      "loss": 0.1634,
      "step": 3283
    },
    {
      "epoch": 4.16,
      "learning_rate": 2.0307414104882462e-05,
      "loss": 0.082,
      "step": 3284
    },
    {
      "epoch": 4.16,
      "learning_rate": 2.029837251356239e-05,
      "loss": 0.0028,
      "step": 3285
    },
    {
      "epoch": 4.16,
      "learning_rate": 2.028933092224232e-05,
      "loss": 0.0018,
      "step": 3286
    },
    {
      "epoch": 4.16,
      "learning_rate": 2.0280289330922243e-05,
      "loss": 0.0019,
      "step": 3287
    },
    {
      "epoch": 4.16,
      "learning_rate": 2.027124773960217e-05,
      "loss": 0.0016,
      "step": 3288
    },
    {
      "epoch": 4.16,
      "learning_rate": 2.0262206148282096e-05,
      "loss": 0.0019,
      "step": 3289
    },
    {
      "epoch": 4.16,
      "learning_rate": 2.0253164556962025e-05,
      "loss": 0.0046,
      "step": 3290
    },
    {
      "epoch": 4.17,
      "learning_rate": 2.0244122965641953e-05,
      "loss": 0.1758,
      "step": 3291
    },
    {
      "epoch": 4.17,
      "learning_rate": 2.023508137432188e-05,
      "loss": 0.0011,
      "step": 3292
    },
    {
      "epoch": 4.17,
      "learning_rate": 2.022603978300181e-05,
      "loss": 0.0529,
      "step": 3293
    },
    {
      "epoch": 4.17,
      "learning_rate": 2.0216998191681737e-05,
      "loss": 0.0018,
      "step": 3294
    },
    {
      "epoch": 4.17,
      "learning_rate": 2.0207956600361662e-05,
      "loss": 0.0083,
      "step": 3295
    },
    {
      "epoch": 4.17,
      "learning_rate": 2.019891500904159e-05,
      "loss": 0.0075,
      "step": 3296
    },
    {
      "epoch": 4.17,
      "learning_rate": 2.018987341772152e-05,
      "loss": 0.006,
      "step": 3297
    },
    {
      "epoch": 4.17,
      "learning_rate": 2.0180831826401447e-05,
      "loss": 0.0034,
      "step": 3298
    },
    {
      "epoch": 4.18,
      "learning_rate": 2.0171790235081375e-05,
      "loss": 0.0015,
      "step": 3299
    },
    {
      "epoch": 4.18,
      "learning_rate": 2.0162748643761304e-05,
      "loss": 0.0047,
      "step": 3300
    },
    {
      "epoch": 4.18,
      "learning_rate": 2.0153707052441232e-05,
      "loss": 0.3622,
      "step": 3301
    },
    {
      "epoch": 4.18,
      "learning_rate": 2.014466546112116e-05,
      "loss": 0.0105,
      "step": 3302
    },
    {
      "epoch": 4.18,
      "learning_rate": 2.0135623869801085e-05,
      "loss": 0.0034,
      "step": 3303
    },
    {
      "epoch": 4.18,
      "learning_rate": 2.0126582278481013e-05,
      "loss": 0.0802,
      "step": 3304
    },
    {
      "epoch": 4.18,
      "learning_rate": 2.011754068716094e-05,
      "loss": 0.0188,
      "step": 3305
    },
    {
      "epoch": 4.18,
      "learning_rate": 2.010849909584087e-05,
      "loss": 0.0026,
      "step": 3306
    },
    {
      "epoch": 4.19,
      "learning_rate": 2.0099457504520798e-05,
      "loss": 0.0019,
      "step": 3307
    },
    {
      "epoch": 4.19,
      "learning_rate": 2.0090415913200726e-05,
      "loss": 0.0036,
      "step": 3308
    },
    {
      "epoch": 4.19,
      "learning_rate": 2.0081374321880654e-05,
      "loss": 0.0195,
      "step": 3309
    },
    {
      "epoch": 4.19,
      "learning_rate": 2.007233273056058e-05,
      "loss": 0.0015,
      "step": 3310
    },
    {
      "epoch": 4.19,
      "learning_rate": 2.0063291139240507e-05,
      "loss": 0.0012,
      "step": 3311
    },
    {
      "epoch": 4.19,
      "learning_rate": 2.0054249547920436e-05,
      "loss": 0.0011,
      "step": 3312
    },
    {
      "epoch": 4.19,
      "learning_rate": 2.004520795660036e-05,
      "loss": 0.0057,
      "step": 3313
    },
    {
      "epoch": 4.19,
      "learning_rate": 2.003616636528029e-05,
      "loss": 0.0058,
      "step": 3314
    },
    {
      "epoch": 4.2,
      "learning_rate": 2.0027124773960217e-05,
      "loss": 0.0025,
      "step": 3315
    },
    {
      "epoch": 4.2,
      "learning_rate": 2.0018083182640145e-05,
      "loss": 0.0587,
      "step": 3316
    },
    {
      "epoch": 4.2,
      "learning_rate": 2.0009041591320073e-05,
      "loss": 0.0061,
      "step": 3317
    },
    {
      "epoch": 4.2,
      "learning_rate": 2e-05,
      "loss": 0.003,
      "step": 3318
    },
    {
      "epoch": 4.2,
      "learning_rate": 1.9990958408679926e-05,
      "loss": 0.0044,
      "step": 3319
    },
    {
      "epoch": 4.2,
      "learning_rate": 1.9981916817359855e-05,
      "loss": 0.0043,
      "step": 3320
    },
    {
      "epoch": 4.2,
      "learning_rate": 1.9972875226039783e-05,
      "loss": 0.0009,
      "step": 3321
    },
    {
      "epoch": 4.2,
      "learning_rate": 1.996383363471971e-05,
      "loss": 0.0033,
      "step": 3322
    },
    {
      "epoch": 4.21,
      "learning_rate": 1.995479204339964e-05,
      "loss": 0.0021,
      "step": 3323
    },
    {
      "epoch": 4.21,
      "learning_rate": 1.9945750452079568e-05,
      "loss": 0.0033,
      "step": 3324
    },
    {
      "epoch": 4.21,
      "learning_rate": 1.9936708860759496e-05,
      "loss": 0.0075,
      "step": 3325
    },
    {
      "epoch": 4.21,
      "learning_rate": 1.9927667269439424e-05,
      "loss": 0.0739,
      "step": 3326
    },
    {
      "epoch": 4.21,
      "learning_rate": 1.991862567811935e-05,
      "loss": 0.0556,
      "step": 3327
    },
    {
      "epoch": 4.21,
      "learning_rate": 1.9909584086799277e-05,
      "loss": 0.0088,
      "step": 3328
    },
    {
      "epoch": 4.21,
      "learning_rate": 1.9900542495479205e-05,
      "loss": 0.0023,
      "step": 3329
    },
    {
      "epoch": 4.22,
      "learning_rate": 1.9891500904159134e-05,
      "loss": 0.0803,
      "step": 3330
    },
    {
      "epoch": 4.22,
      "learning_rate": 1.9882459312839062e-05,
      "loss": 0.0306,
      "step": 3331
    },
    {
      "epoch": 4.22,
      "learning_rate": 1.987341772151899e-05,
      "loss": 0.0013,
      "step": 3332
    },
    {
      "epoch": 4.22,
      "learning_rate": 1.986437613019892e-05,
      "loss": 0.0021,
      "step": 3333
    },
    {
      "epoch": 4.22,
      "learning_rate": 1.9855334538878843e-05,
      "loss": 0.001,
      "step": 3334
    },
    {
      "epoch": 4.22,
      "learning_rate": 1.984629294755877e-05,
      "loss": 0.0025,
      "step": 3335
    },
    {
      "epoch": 4.22,
      "learning_rate": 1.9837251356238696e-05,
      "loss": 0.0145,
      "step": 3336
    },
    {
      "epoch": 4.22,
      "learning_rate": 1.9828209764918625e-05,
      "loss": 0.0042,
      "step": 3337
    },
    {
      "epoch": 4.23,
      "learning_rate": 1.9819168173598553e-05,
      "loss": 0.0045,
      "step": 3338
    },
    {
      "epoch": 4.23,
      "learning_rate": 1.981012658227848e-05,
      "loss": 0.0014,
      "step": 3339
    },
    {
      "epoch": 4.23,
      "learning_rate": 1.980108499095841e-05,
      "loss": 0.0007,
      "step": 3340
    },
    {
      "epoch": 4.23,
      "learning_rate": 1.9792043399638338e-05,
      "loss": 0.0156,
      "step": 3341
    },
    {
      "epoch": 4.23,
      "learning_rate": 1.9783001808318266e-05,
      "loss": 0.0009,
      "step": 3342
    },
    {
      "epoch": 4.23,
      "learning_rate": 1.977396021699819e-05,
      "loss": 0.1331,
      "step": 3343
    },
    {
      "epoch": 4.23,
      "learning_rate": 1.976491862567812e-05,
      "loss": 0.1671,
      "step": 3344
    },
    {
      "epoch": 4.23,
      "learning_rate": 1.9755877034358047e-05,
      "loss": 0.1905,
      "step": 3345
    },
    {
      "epoch": 4.24,
      "learning_rate": 1.9746835443037975e-05,
      "loss": 0.0023,
      "step": 3346
    },
    {
      "epoch": 4.24,
      "learning_rate": 1.9737793851717904e-05,
      "loss": 0.0334,
      "step": 3347
    },
    {
      "epoch": 4.24,
      "learning_rate": 1.9728752260397832e-05,
      "loss": 0.0005,
      "step": 3348
    },
    {
      "epoch": 4.24,
      "learning_rate": 1.971971066907776e-05,
      "loss": 0.0023,
      "step": 3349
    },
    {
      "epoch": 4.24,
      "learning_rate": 1.971066907775769e-05,
      "loss": 0.0118,
      "step": 3350
    },
    {
      "epoch": 4.24,
      "learning_rate": 1.9701627486437613e-05,
      "loss": 0.0023,
      "step": 3351
    },
    {
      "epoch": 4.24,
      "learning_rate": 1.969258589511754e-05,
      "loss": 0.0098,
      "step": 3352
    },
    {
      "epoch": 4.24,
      "learning_rate": 1.968354430379747e-05,
      "loss": 0.0543,
      "step": 3353
    },
    {
      "epoch": 4.25,
      "learning_rate": 1.9674502712477398e-05,
      "loss": 0.0226,
      "step": 3354
    },
    {
      "epoch": 4.25,
      "learning_rate": 1.9665461121157326e-05,
      "loss": 0.1704,
      "step": 3355
    },
    {
      "epoch": 4.25,
      "learning_rate": 1.9656419529837254e-05,
      "loss": 0.0157,
      "step": 3356
    },
    {
      "epoch": 4.25,
      "learning_rate": 1.964737793851718e-05,
      "loss": 0.0026,
      "step": 3357
    },
    {
      "epoch": 4.25,
      "learning_rate": 1.9638336347197107e-05,
      "loss": 0.1293,
      "step": 3358
    },
    {
      "epoch": 4.25,
      "learning_rate": 1.9629294755877036e-05,
      "loss": 0.0204,
      "step": 3359
    },
    {
      "epoch": 4.25,
      "learning_rate": 1.962025316455696e-05,
      "loss": 0.016,
      "step": 3360
    },
    {
      "epoch": 4.25,
      "learning_rate": 1.961121157323689e-05,
      "loss": 0.0025,
      "step": 3361
    },
    {
      "epoch": 4.26,
      "learning_rate": 1.9602169981916817e-05,
      "loss": 0.0513,
      "step": 3362
    },
    {
      "epoch": 4.26,
      "learning_rate": 1.9593128390596745e-05,
      "loss": 0.0008,
      "step": 3363
    },
    {
      "epoch": 4.26,
      "learning_rate": 1.9584086799276673e-05,
      "loss": 0.0056,
      "step": 3364
    },
    {
      "epoch": 4.26,
      "learning_rate": 1.9575045207956602e-05,
      "loss": 0.0081,
      "step": 3365
    },
    {
      "epoch": 4.26,
      "learning_rate": 1.956600361663653e-05,
      "loss": 0.1183,
      "step": 3366
    },
    {
      "epoch": 4.26,
      "learning_rate": 1.9556962025316455e-05,
      "loss": 0.0072,
      "step": 3367
    },
    {
      "epoch": 4.26,
      "learning_rate": 1.9547920433996383e-05,
      "loss": 0.0045,
      "step": 3368
    },
    {
      "epoch": 4.26,
      "learning_rate": 1.953887884267631e-05,
      "loss": 0.0056,
      "step": 3369
    },
    {
      "epoch": 4.27,
      "learning_rate": 1.952983725135624e-05,
      "loss": 0.054,
      "step": 3370
    },
    {
      "epoch": 4.27,
      "learning_rate": 1.9520795660036168e-05,
      "loss": 0.0038,
      "step": 3371
    },
    {
      "epoch": 4.27,
      "learning_rate": 1.9511754068716096e-05,
      "loss": 0.0017,
      "step": 3372
    },
    {
      "epoch": 4.27,
      "learning_rate": 1.9502712477396024e-05,
      "loss": 0.0423,
      "step": 3373
    },
    {
      "epoch": 4.27,
      "learning_rate": 1.9493670886075952e-05,
      "loss": 0.0033,
      "step": 3374
    },
    {
      "epoch": 4.27,
      "learning_rate": 1.9484629294755877e-05,
      "loss": 0.0024,
      "step": 3375
    },
    {
      "epoch": 4.27,
      "learning_rate": 1.9475587703435806e-05,
      "loss": 0.0086,
      "step": 3376
    },
    {
      "epoch": 4.27,
      "learning_rate": 1.9466546112115734e-05,
      "loss": 0.0043,
      "step": 3377
    },
    {
      "epoch": 4.28,
      "learning_rate": 1.9457504520795662e-05,
      "loss": 0.0717,
      "step": 3378
    },
    {
      "epoch": 4.28,
      "learning_rate": 1.944846292947559e-05,
      "loss": 0.0022,
      "step": 3379
    },
    {
      "epoch": 4.28,
      "learning_rate": 1.9439421338155515e-05,
      "loss": 0.0028,
      "step": 3380
    },
    {
      "epoch": 4.28,
      "learning_rate": 1.9430379746835443e-05,
      "loss": 0.0413,
      "step": 3381
    },
    {
      "epoch": 4.28,
      "learning_rate": 1.942133815551537e-05,
      "loss": 0.0019,
      "step": 3382
    },
    {
      "epoch": 4.28,
      "learning_rate": 1.9412296564195296e-05,
      "loss": 0.0033,
      "step": 3383
    },
    {
      "epoch": 4.28,
      "learning_rate": 1.9403254972875225e-05,
      "loss": 0.1299,
      "step": 3384
    },
    {
      "epoch": 4.28,
      "learning_rate": 1.9394213381555153e-05,
      "loss": 0.006,
      "step": 3385
    },
    {
      "epoch": 4.29,
      "learning_rate": 1.938517179023508e-05,
      "loss": 0.0013,
      "step": 3386
    },
    {
      "epoch": 4.29,
      "learning_rate": 1.937613019891501e-05,
      "loss": 0.001,
      "step": 3387
    },
    {
      "epoch": 4.29,
      "learning_rate": 1.9367088607594938e-05,
      "loss": 0.009,
      "step": 3388
    },
    {
      "epoch": 4.29,
      "learning_rate": 1.9358047016274866e-05,
      "loss": 0.001,
      "step": 3389
    },
    {
      "epoch": 4.29,
      "learning_rate": 1.9349005424954794e-05,
      "loss": 0.002,
      "step": 3390
    },
    {
      "epoch": 4.29,
      "learning_rate": 1.933996383363472e-05,
      "loss": 0.0888,
      "step": 3391
    },
    {
      "epoch": 4.29,
      "learning_rate": 1.9330922242314647e-05,
      "loss": 0.0027,
      "step": 3392
    },
    {
      "epoch": 4.29,
      "learning_rate": 1.9321880650994575e-05,
      "loss": 0.0033,
      "step": 3393
    },
    {
      "epoch": 4.3,
      "learning_rate": 1.9312839059674504e-05,
      "loss": 0.0196,
      "step": 3394
    },
    {
      "epoch": 4.3,
      "learning_rate": 1.9303797468354432e-05,
      "loss": 0.0305,
      "step": 3395
    },
    {
      "epoch": 4.3,
      "learning_rate": 1.929475587703436e-05,
      "loss": 0.0035,
      "step": 3396
    },
    {
      "epoch": 4.3,
      "learning_rate": 1.928571428571429e-05,
      "loss": 0.0014,
      "step": 3397
    },
    {
      "epoch": 4.3,
      "learning_rate": 1.9276672694394217e-05,
      "loss": 0.0021,
      "step": 3398
    },
    {
      "epoch": 4.3,
      "learning_rate": 1.926763110307414e-05,
      "loss": 0.0013,
      "step": 3399
    },
    {
      "epoch": 4.3,
      "learning_rate": 1.925858951175407e-05,
      "loss": 0.0046,
      "step": 3400
    },
    {
      "epoch": 4.3,
      "learning_rate": 1.9249547920433998e-05,
      "loss": 0.0097,
      "step": 3401
    },
    {
      "epoch": 4.31,
      "learning_rate": 1.9240506329113926e-05,
      "loss": 0.0017,
      "step": 3402
    },
    {
      "epoch": 4.31,
      "learning_rate": 1.9231464737793854e-05,
      "loss": 0.0019,
      "step": 3403
    },
    {
      "epoch": 4.31,
      "learning_rate": 1.922242314647378e-05,
      "loss": 0.0036,
      "step": 3404
    },
    {
      "epoch": 4.31,
      "learning_rate": 1.9213381555153708e-05,
      "loss": 0.0011,
      "step": 3405
    },
    {
      "epoch": 4.31,
      "learning_rate": 1.9204339963833636e-05,
      "loss": 0.0054,
      "step": 3406
    },
    {
      "epoch": 4.31,
      "learning_rate": 1.919529837251356e-05,
      "loss": 0.0035,
      "step": 3407
    },
    {
      "epoch": 4.31,
      "learning_rate": 1.918625678119349e-05,
      "loss": 0.0007,
      "step": 3408
    },
    {
      "epoch": 4.31,
      "learning_rate": 1.9177215189873417e-05,
      "loss": 0.1465,
      "step": 3409
    },
    {
      "epoch": 4.32,
      "learning_rate": 1.9168173598553345e-05,
      "loss": 0.1952,
      "step": 3410
    },
    {
      "epoch": 4.32,
      "learning_rate": 1.9159132007233274e-05,
      "loss": 0.0008,
      "step": 3411
    },
    {
      "epoch": 4.32,
      "learning_rate": 1.9150090415913202e-05,
      "loss": 0.0015,
      "step": 3412
    },
    {
      "epoch": 4.32,
      "learning_rate": 1.914104882459313e-05,
      "loss": 0.0012,
      "step": 3413
    },
    {
      "epoch": 4.32,
      "learning_rate": 1.9132007233273058e-05,
      "loss": 0.003,
      "step": 3414
    },
    {
      "epoch": 4.32,
      "learning_rate": 1.9122965641952983e-05,
      "loss": 0.0053,
      "step": 3415
    },
    {
      "epoch": 4.32,
      "learning_rate": 1.911392405063291e-05,
      "loss": 0.0047,
      "step": 3416
    },
    {
      "epoch": 4.33,
      "learning_rate": 1.910488245931284e-05,
      "loss": 0.0006,
      "step": 3417
    },
    {
      "epoch": 4.33,
      "learning_rate": 1.9095840867992768e-05,
      "loss": 0.0007,
      "step": 3418
    },
    {
      "epoch": 4.33,
      "learning_rate": 1.9086799276672696e-05,
      "loss": 0.0013,
      "step": 3419
    },
    {
      "epoch": 4.33,
      "learning_rate": 1.9077757685352624e-05,
      "loss": 0.0021,
      "step": 3420
    },
    {
      "epoch": 4.33,
      "learning_rate": 1.9068716094032553e-05,
      "loss": 0.0609,
      "step": 3421
    },
    {
      "epoch": 4.33,
      "learning_rate": 1.905967450271248e-05,
      "loss": 0.0047,
      "step": 3422
    },
    {
      "epoch": 4.33,
      "learning_rate": 1.9050632911392406e-05,
      "loss": 0.0014,
      "step": 3423
    },
    {
      "epoch": 4.33,
      "learning_rate": 1.9041591320072334e-05,
      "loss": 0.001,
      "step": 3424
    },
    {
      "epoch": 4.34,
      "learning_rate": 1.9032549728752262e-05,
      "loss": 0.0023,
      "step": 3425
    },
    {
      "epoch": 4.34,
      "learning_rate": 1.902350813743219e-05,
      "loss": 0.1987,
      "step": 3426
    },
    {
      "epoch": 4.34,
      "learning_rate": 1.9014466546112115e-05,
      "loss": 0.0004,
      "step": 3427
    },
    {
      "epoch": 4.34,
      "learning_rate": 1.9005424954792043e-05,
      "loss": 0.0008,
      "step": 3428
    },
    {
      "epoch": 4.34,
      "learning_rate": 1.899638336347197e-05,
      "loss": 0.0024,
      "step": 3429
    },
    {
      "epoch": 4.34,
      "learning_rate": 1.89873417721519e-05,
      "loss": 0.0028,
      "step": 3430
    },
    {
      "epoch": 4.34,
      "learning_rate": 1.8978300180831825e-05,
      "loss": 0.1074,
      "step": 3431
    },
    {
      "epoch": 4.34,
      "learning_rate": 1.8969258589511753e-05,
      "loss": 0.0112,
      "step": 3432
    },
    {
      "epoch": 4.35,
      "learning_rate": 1.896021699819168e-05,
      "loss": 0.0022,
      "step": 3433
    },
    {
      "epoch": 4.35,
      "learning_rate": 1.895117540687161e-05,
      "loss": 0.0005,
      "step": 3434
    },
    {
      "epoch": 4.35,
      "learning_rate": 1.8942133815551538e-05,
      "loss": 0.0009,
      "step": 3435
    },
    {
      "epoch": 4.35,
      "learning_rate": 1.8933092224231466e-05,
      "loss": 0.0011,
      "step": 3436
    },
    {
      "epoch": 4.35,
      "learning_rate": 1.8924050632911394e-05,
      "loss": 0.0008,
      "step": 3437
    },
    {
      "epoch": 4.35,
      "learning_rate": 1.8915009041591322e-05,
      "loss": 0.0068,
      "step": 3438
    },
    {
      "epoch": 4.35,
      "learning_rate": 1.8905967450271247e-05,
      "loss": 0.0018,
      "step": 3439
    },
    {
      "epoch": 4.35,
      "learning_rate": 1.8896925858951176e-05,
      "loss": 0.0008,
      "step": 3440
    },
    {
      "epoch": 4.36,
      "learning_rate": 1.8887884267631104e-05,
      "loss": 0.0065,
      "step": 3441
    },
    {
      "epoch": 4.36,
      "learning_rate": 1.8878842676311032e-05,
      "loss": 0.0374,
      "step": 3442
    },
    {
      "epoch": 4.36,
      "learning_rate": 1.886980108499096e-05,
      "loss": 0.0026,
      "step": 3443
    },
    {
      "epoch": 4.36,
      "learning_rate": 1.886075949367089e-05,
      "loss": 0.1085,
      "step": 3444
    },
    {
      "epoch": 4.36,
      "learning_rate": 1.8851717902350817e-05,
      "loss": 0.001,
      "step": 3445
    },
    {
      "epoch": 4.36,
      "learning_rate": 1.8842676311030745e-05,
      "loss": 0.0004,
      "step": 3446
    },
    {
      "epoch": 4.36,
      "learning_rate": 1.883363471971067e-05,
      "loss": 0.0044,
      "step": 3447
    },
    {
      "epoch": 4.36,
      "learning_rate": 1.8824593128390598e-05,
      "loss": 0.0014,
      "step": 3448
    },
    {
      "epoch": 4.37,
      "learning_rate": 1.8815551537070526e-05,
      "loss": 0.0004,
      "step": 3449
    },
    {
      "epoch": 4.37,
      "learning_rate": 1.8806509945750454e-05,
      "loss": 0.0299,
      "step": 3450
    },
    {
      "epoch": 4.37,
      "learning_rate": 1.879746835443038e-05,
      "loss": 0.0043,
      "step": 3451
    },
    {
      "epoch": 4.37,
      "learning_rate": 1.8788426763110308e-05,
      "loss": 0.1023,
      "step": 3452
    },
    {
      "epoch": 4.37,
      "learning_rate": 1.8779385171790236e-05,
      "loss": 0.0016,
      "step": 3453
    },
    {
      "epoch": 4.37,
      "learning_rate": 1.8770343580470164e-05,
      "loss": 0.0029,
      "step": 3454
    },
    {
      "epoch": 4.37,
      "learning_rate": 1.876130198915009e-05,
      "loss": 0.0004,
      "step": 3455
    },
    {
      "epoch": 4.37,
      "learning_rate": 1.8752260397830017e-05,
      "loss": 0.0316,
      "step": 3456
    },
    {
      "epoch": 4.38,
      "learning_rate": 1.8743218806509945e-05,
      "loss": 0.0054,
      "step": 3457
    },
    {
      "epoch": 4.38,
      "learning_rate": 1.8734177215189874e-05,
      "loss": 0.0014,
      "step": 3458
    },
    {
      "epoch": 4.38,
      "learning_rate": 1.8725135623869802e-05,
      "loss": 0.151,
      "step": 3459
    },
    {
      "epoch": 4.38,
      "learning_rate": 1.871609403254973e-05,
      "loss": 0.0155,
      "step": 3460
    },
    {
      "epoch": 4.38,
      "learning_rate": 1.870705244122966e-05,
      "loss": 0.1194,
      "step": 3461
    },
    {
      "epoch": 4.38,
      "learning_rate": 1.8698010849909587e-05,
      "loss": 0.0008,
      "step": 3462
    },
    {
      "epoch": 4.38,
      "learning_rate": 1.868896925858951e-05,
      "loss": 0.0788,
      "step": 3463
    },
    {
      "epoch": 4.38,
      "learning_rate": 1.867992766726944e-05,
      "loss": 0.0005,
      "step": 3464
    },
    {
      "epoch": 4.39,
      "learning_rate": 1.8670886075949368e-05,
      "loss": 0.0006,
      "step": 3465
    },
    {
      "epoch": 4.39,
      "learning_rate": 1.8661844484629296e-05,
      "loss": 0.0018,
      "step": 3466
    },
    {
      "epoch": 4.39,
      "learning_rate": 1.8652802893309224e-05,
      "loss": 0.0447,
      "step": 3467
    },
    {
      "epoch": 4.39,
      "learning_rate": 1.8643761301989153e-05,
      "loss": 0.0063,
      "step": 3468
    },
    {
      "epoch": 4.39,
      "learning_rate": 1.863471971066908e-05,
      "loss": 0.0702,
      "step": 3469
    },
    {
      "epoch": 4.39,
      "learning_rate": 1.862567811934901e-05,
      "loss": 0.0606,
      "step": 3470
    },
    {
      "epoch": 4.39,
      "learning_rate": 1.8616636528028934e-05,
      "loss": 0.0056,
      "step": 3471
    },
    {
      "epoch": 4.39,
      "learning_rate": 1.8607594936708862e-05,
      "loss": 0.001,
      "step": 3472
    },
    {
      "epoch": 4.4,
      "learning_rate": 1.859855334538879e-05,
      "loss": 0.2754,
      "step": 3473
    },
    {
      "epoch": 4.4,
      "learning_rate": 1.8589511754068715e-05,
      "loss": 0.001,
      "step": 3474
    },
    {
      "epoch": 4.4,
      "learning_rate": 1.8580470162748643e-05,
      "loss": 0.002,
      "step": 3475
    },
    {
      "epoch": 4.4,
      "learning_rate": 1.8571428571428572e-05,
      "loss": 0.0022,
      "step": 3476
    },
    {
      "epoch": 4.4,
      "learning_rate": 1.85623869801085e-05,
      "loss": 0.0008,
      "step": 3477
    },
    {
      "epoch": 4.4,
      "learning_rate": 1.8553345388788428e-05,
      "loss": 0.1144,
      "step": 3478
    },
    {
      "epoch": 4.4,
      "learning_rate": 1.8544303797468353e-05,
      "loss": 0.0011,
      "step": 3479
    },
    {
      "epoch": 4.4,
      "learning_rate": 1.853526220614828e-05,
      "loss": 0.0024,
      "step": 3480
    },
    {
      "epoch": 4.41,
      "learning_rate": 1.852622061482821e-05,
      "loss": 0.0015,
      "step": 3481
    },
    {
      "epoch": 4.41,
      "learning_rate": 1.8517179023508138e-05,
      "loss": 0.0159,
      "step": 3482
    },
    {
      "epoch": 4.41,
      "learning_rate": 1.8508137432188066e-05,
      "loss": 0.0073,
      "step": 3483
    },
    {
      "epoch": 4.41,
      "learning_rate": 1.8499095840867994e-05,
      "loss": 0.0018,
      "step": 3484
    },
    {
      "epoch": 4.41,
      "learning_rate": 1.8490054249547922e-05,
      "loss": 0.0013,
      "step": 3485
    },
    {
      "epoch": 4.41,
      "learning_rate": 1.848101265822785e-05,
      "loss": 0.0032,
      "step": 3486
    },
    {
      "epoch": 4.41,
      "learning_rate": 1.8471971066907776e-05,
      "loss": 0.0007,
      "step": 3487
    },
    {
      "epoch": 4.41,
      "learning_rate": 1.8462929475587704e-05,
      "loss": 0.0366,
      "step": 3488
    },
    {
      "epoch": 4.42,
      "learning_rate": 1.8453887884267632e-05,
      "loss": 0.0012,
      "step": 3489
    },
    {
      "epoch": 4.42,
      "learning_rate": 1.844484629294756e-05,
      "loss": 0.0015,
      "step": 3490
    },
    {
      "epoch": 4.42,
      "learning_rate": 1.843580470162749e-05,
      "loss": 0.0011,
      "step": 3491
    },
    {
      "epoch": 4.42,
      "learning_rate": 1.8426763110307417e-05,
      "loss": 0.0015,
      "step": 3492
    },
    {
      "epoch": 4.42,
      "learning_rate": 1.8417721518987345e-05,
      "loss": 0.0015,
      "step": 3493
    },
    {
      "epoch": 4.42,
      "learning_rate": 1.8408679927667273e-05,
      "loss": 0.0007,
      "step": 3494
    },
    {
      "epoch": 4.42,
      "learning_rate": 1.8399638336347198e-05,
      "loss": 0.1,
      "step": 3495
    },
    {
      "epoch": 4.43,
      "learning_rate": 1.8390596745027126e-05,
      "loss": 0.0006,
      "step": 3496
    },
    {
      "epoch": 4.43,
      "learning_rate": 1.8381555153707055e-05,
      "loss": 0.1854,
      "step": 3497
    },
    {
      "epoch": 4.43,
      "learning_rate": 1.837251356238698e-05,
      "loss": 0.0009,
      "step": 3498
    },
    {
      "epoch": 4.43,
      "learning_rate": 1.8363471971066908e-05,
      "loss": 0.0053,
      "step": 3499
    },
    {
      "epoch": 4.43,
      "learning_rate": 1.8354430379746836e-05,
      "loss": 0.0009,
      "step": 3500
    },
    {
      "epoch": 4.43,
      "eval_accuracy": 0.9084084084084084,
      "eval_loss": 0.45180293917655945,
      "eval_runtime": 45.1565,
      "eval_samples_per_second": 29.497,
      "eval_steps_per_second": 3.698,
      "step": 3500
    },
    {
      "epoch": 4.43,
      "learning_rate": 1.8345388788426764e-05,
      "loss": 0.0009,
      "step": 3501
    },
    {
      "epoch": 4.43,
      "learning_rate": 1.8336347197106692e-05,
      "loss": 0.0057,
      "step": 3502
    },
    {
      "epoch": 4.43,
      "learning_rate": 1.8327305605786617e-05,
      "loss": 0.0009,
      "step": 3503
    },
    {
      "epoch": 4.44,
      "learning_rate": 1.8318264014466545e-05,
      "loss": 0.0565,
      "step": 3504
    },
    {
      "epoch": 4.44,
      "learning_rate": 1.8309222423146474e-05,
      "loss": 0.1435,
      "step": 3505
    },
    {
      "epoch": 4.44,
      "learning_rate": 1.8300180831826402e-05,
      "loss": 0.0009,
      "step": 3506
    },
    {
      "epoch": 4.44,
      "learning_rate": 1.829113924050633e-05,
      "loss": 0.0041,
      "step": 3507
    },
    {
      "epoch": 4.44,
      "learning_rate": 1.828209764918626e-05,
      "loss": 0.093,
      "step": 3508
    },
    {
      "epoch": 4.44,
      "learning_rate": 1.8273056057866187e-05,
      "loss": 0.001,
      "step": 3509
    },
    {
      "epoch": 4.44,
      "learning_rate": 1.8264014466546115e-05,
      "loss": 0.0915,
      "step": 3510
    },
    {
      "epoch": 4.44,
      "learning_rate": 1.825497287522604e-05,
      "loss": 0.0068,
      "step": 3511
    },
    {
      "epoch": 4.45,
      "learning_rate": 1.8245931283905968e-05,
      "loss": 0.0013,
      "step": 3512
    },
    {
      "epoch": 4.45,
      "learning_rate": 1.8236889692585896e-05,
      "loss": 0.0018,
      "step": 3513
    },
    {
      "epoch": 4.45,
      "learning_rate": 1.8227848101265824e-05,
      "loss": 0.2451,
      "step": 3514
    },
    {
      "epoch": 4.45,
      "learning_rate": 1.8218806509945753e-05,
      "loss": 0.004,
      "step": 3515
    },
    {
      "epoch": 4.45,
      "learning_rate": 1.820976491862568e-05,
      "loss": 0.0009,
      "step": 3516
    },
    {
      "epoch": 4.45,
      "learning_rate": 1.820072332730561e-05,
      "loss": 0.0005,
      "step": 3517
    },
    {
      "epoch": 4.45,
      "learning_rate": 1.8191681735985537e-05,
      "loss": 0.0012,
      "step": 3518
    },
    {
      "epoch": 4.45,
      "learning_rate": 1.8182640144665462e-05,
      "loss": 0.0008,
      "step": 3519
    },
    {
      "epoch": 4.46,
      "learning_rate": 1.817359855334539e-05,
      "loss": 0.0016,
      "step": 3520
    },
    {
      "epoch": 4.46,
      "learning_rate": 1.8164556962025315e-05,
      "loss": 0.2095,
      "step": 3521
    },
    {
      "epoch": 4.46,
      "learning_rate": 1.8155515370705244e-05,
      "loss": 0.0008,
      "step": 3522
    },
    {
      "epoch": 4.46,
      "learning_rate": 1.8146473779385172e-05,
      "loss": 0.0046,
      "step": 3523
    },
    {
      "epoch": 4.46,
      "learning_rate": 1.81374321880651e-05,
      "loss": 0.0004,
      "step": 3524
    },
    {
      "epoch": 4.46,
      "learning_rate": 1.8128390596745028e-05,
      "loss": 0.0081,
      "step": 3525
    },
    {
      "epoch": 4.46,
      "learning_rate": 1.8119349005424953e-05,
      "loss": 0.0007,
      "step": 3526
    },
    {
      "epoch": 4.46,
      "learning_rate": 1.811030741410488e-05,
      "loss": 0.0042,
      "step": 3527
    },
    {
      "epoch": 4.47,
      "learning_rate": 1.810126582278481e-05,
      "loss": 0.0017,
      "step": 3528
    },
    {
      "epoch": 4.47,
      "learning_rate": 1.8092224231464738e-05,
      "loss": 0.0042,
      "step": 3529
    },
    {
      "epoch": 4.47,
      "learning_rate": 1.8083182640144666e-05,
      "loss": 0.0006,
      "step": 3530
    },
    {
      "epoch": 4.47,
      "learning_rate": 1.8074141048824594e-05,
      "loss": 0.2531,
      "step": 3531
    },
    {
      "epoch": 4.47,
      "learning_rate": 1.8065099457504523e-05,
      "loss": 0.0008,
      "step": 3532
    },
    {
      "epoch": 4.47,
      "learning_rate": 1.805605786618445e-05,
      "loss": 0.0005,
      "step": 3533
    },
    {
      "epoch": 4.47,
      "learning_rate": 1.8047016274864376e-05,
      "loss": 0.0278,
      "step": 3534
    },
    {
      "epoch": 4.47,
      "learning_rate": 1.8037974683544304e-05,
      "loss": 0.0006,
      "step": 3535
    },
    {
      "epoch": 4.48,
      "learning_rate": 1.8028933092224232e-05,
      "loss": 0.0011,
      "step": 3536
    },
    {
      "epoch": 4.48,
      "learning_rate": 1.801989150090416e-05,
      "loss": 0.0097,
      "step": 3537
    },
    {
      "epoch": 4.48,
      "learning_rate": 1.801084990958409e-05,
      "loss": 0.0036,
      "step": 3538
    },
    {
      "epoch": 4.48,
      "learning_rate": 1.8001808318264017e-05,
      "loss": 0.0013,
      "step": 3539
    },
    {
      "epoch": 4.48,
      "learning_rate": 1.7992766726943945e-05,
      "loss": 0.1147,
      "step": 3540
    },
    {
      "epoch": 4.48,
      "learning_rate": 1.7983725135623873e-05,
      "loss": 0.0017,
      "step": 3541
    },
    {
      "epoch": 4.48,
      "learning_rate": 1.7974683544303798e-05,
      "loss": 0.0007,
      "step": 3542
    },
    {
      "epoch": 4.48,
      "learning_rate": 1.7965641952983726e-05,
      "loss": 0.0011,
      "step": 3543
    },
    {
      "epoch": 4.49,
      "learning_rate": 1.7956600361663655e-05,
      "loss": 0.0031,
      "step": 3544
    },
    {
      "epoch": 4.49,
      "learning_rate": 1.794755877034358e-05,
      "loss": 0.0006,
      "step": 3545
    },
    {
      "epoch": 4.49,
      "learning_rate": 1.7938517179023508e-05,
      "loss": 0.0054,
      "step": 3546
    },
    {
      "epoch": 4.49,
      "learning_rate": 1.7929475587703436e-05,
      "loss": 0.0007,
      "step": 3547
    },
    {
      "epoch": 4.49,
      "learning_rate": 1.7920433996383364e-05,
      "loss": 0.0022,
      "step": 3548
    },
    {
      "epoch": 4.49,
      "learning_rate": 1.7911392405063292e-05,
      "loss": 0.0337,
      "step": 3549
    },
    {
      "epoch": 4.49,
      "learning_rate": 1.7902350813743217e-05,
      "loss": 0.0007,
      "step": 3550
    },
    {
      "epoch": 4.49,
      "learning_rate": 1.7893309222423146e-05,
      "loss": 0.0012,
      "step": 3551
    },
    {
      "epoch": 4.5,
      "learning_rate": 1.7884267631103074e-05,
      "loss": 0.0022,
      "step": 3552
    },
    {
      "epoch": 4.5,
      "learning_rate": 1.7875226039783002e-05,
      "loss": 0.0006,
      "step": 3553
    },
    {
      "epoch": 4.5,
      "learning_rate": 1.786618444846293e-05,
      "loss": 0.0009,
      "step": 3554
    },
    {
      "epoch": 4.5,
      "learning_rate": 1.785714285714286e-05,
      "loss": 0.0007,
      "step": 3555
    },
    {
      "epoch": 4.5,
      "learning_rate": 1.7848101265822787e-05,
      "loss": 0.2272,
      "step": 3556
    },
    {
      "epoch": 4.5,
      "learning_rate": 1.7839059674502715e-05,
      "loss": 0.0008,
      "step": 3557
    },
    {
      "epoch": 4.5,
      "learning_rate": 1.783001808318264e-05,
      "loss": 0.0008,
      "step": 3558
    },
    {
      "epoch": 4.5,
      "learning_rate": 1.7820976491862568e-05,
      "loss": 0.0101,
      "step": 3559
    },
    {
      "epoch": 4.51,
      "learning_rate": 1.7811934900542496e-05,
      "loss": 0.0008,
      "step": 3560
    },
    {
      "epoch": 4.51,
      "learning_rate": 1.7802893309222425e-05,
      "loss": 0.0049,
      "step": 3561
    },
    {
      "epoch": 4.51,
      "learning_rate": 1.7793851717902353e-05,
      "loss": 0.2965,
      "step": 3562
    },
    {
      "epoch": 4.51,
      "learning_rate": 1.778481012658228e-05,
      "loss": 0.0178,
      "step": 3563
    },
    {
      "epoch": 4.51,
      "learning_rate": 1.777576853526221e-05,
      "loss": 0.002,
      "step": 3564
    },
    {
      "epoch": 4.51,
      "learning_rate": 1.7766726943942134e-05,
      "loss": 0.0011,
      "step": 3565
    },
    {
      "epoch": 4.51,
      "learning_rate": 1.7757685352622062e-05,
      "loss": 0.0821,
      "step": 3566
    },
    {
      "epoch": 4.51,
      "learning_rate": 1.774864376130199e-05,
      "loss": 0.0006,
      "step": 3567
    },
    {
      "epoch": 4.52,
      "learning_rate": 1.7739602169981915e-05,
      "loss": 0.0012,
      "step": 3568
    },
    {
      "epoch": 4.52,
      "learning_rate": 1.7730560578661844e-05,
      "loss": 0.0015,
      "step": 3569
    },
    {
      "epoch": 4.52,
      "learning_rate": 1.7721518987341772e-05,
      "loss": 0.1302,
      "step": 3570
    },
    {
      "epoch": 4.52,
      "learning_rate": 1.77124773960217e-05,
      "loss": 0.0729,
      "step": 3571
    },
    {
      "epoch": 4.52,
      "learning_rate": 1.770343580470163e-05,
      "loss": 0.002,
      "step": 3572
    },
    {
      "epoch": 4.52,
      "learning_rate": 1.7694394213381557e-05,
      "loss": 0.0056,
      "step": 3573
    },
    {
      "epoch": 4.52,
      "learning_rate": 1.768535262206148e-05,
      "loss": 0.0024,
      "step": 3574
    },
    {
      "epoch": 4.52,
      "learning_rate": 1.767631103074141e-05,
      "loss": 0.0029,
      "step": 3575
    },
    {
      "epoch": 4.53,
      "learning_rate": 1.7667269439421338e-05,
      "loss": 0.001,
      "step": 3576
    },
    {
      "epoch": 4.53,
      "learning_rate": 1.7658227848101266e-05,
      "loss": 0.0008,
      "step": 3577
    },
    {
      "epoch": 4.53,
      "learning_rate": 1.7649186256781194e-05,
      "loss": 0.001,
      "step": 3578
    },
    {
      "epoch": 4.53,
      "learning_rate": 1.7640144665461123e-05,
      "loss": 0.1353,
      "step": 3579
    },
    {
      "epoch": 4.53,
      "learning_rate": 1.763110307414105e-05,
      "loss": 0.0007,
      "step": 3580
    },
    {
      "epoch": 4.53,
      "learning_rate": 1.762206148282098e-05,
      "loss": 0.0286,
      "step": 3581
    },
    {
      "epoch": 4.53,
      "learning_rate": 1.7613019891500904e-05,
      "loss": 0.0017,
      "step": 3582
    },
    {
      "epoch": 4.54,
      "learning_rate": 1.7603978300180832e-05,
      "loss": 0.0306,
      "step": 3583
    },
    {
      "epoch": 4.54,
      "learning_rate": 1.759493670886076e-05,
      "loss": 0.0103,
      "step": 3584
    },
    {
      "epoch": 4.54,
      "learning_rate": 1.758589511754069e-05,
      "loss": 0.0016,
      "step": 3585
    },
    {
      "epoch": 4.54,
      "learning_rate": 1.7576853526220617e-05,
      "loss": 0.0121,
      "step": 3586
    },
    {
      "epoch": 4.54,
      "learning_rate": 1.7567811934900545e-05,
      "loss": 0.0004,
      "step": 3587
    },
    {
      "epoch": 4.54,
      "learning_rate": 1.7558770343580473e-05,
      "loss": 0.0009,
      "step": 3588
    },
    {
      "epoch": 4.54,
      "learning_rate": 1.7549728752260398e-05,
      "loss": 0.001,
      "step": 3589
    },
    {
      "epoch": 4.54,
      "learning_rate": 1.7540687160940326e-05,
      "loss": 0.0341,
      "step": 3590
    },
    {
      "epoch": 4.55,
      "learning_rate": 1.7531645569620255e-05,
      "loss": 0.1797,
      "step": 3591
    },
    {
      "epoch": 4.55,
      "learning_rate": 1.752260397830018e-05,
      "loss": 0.0009,
      "step": 3592
    },
    {
      "epoch": 4.55,
      "learning_rate": 1.7513562386980108e-05,
      "loss": 0.0019,
      "step": 3593
    },
    {
      "epoch": 4.55,
      "learning_rate": 1.7504520795660036e-05,
      "loss": 0.0011,
      "step": 3594
    },
    {
      "epoch": 4.55,
      "learning_rate": 1.7495479204339964e-05,
      "loss": 0.0009,
      "step": 3595
    },
    {
      "epoch": 4.55,
      "learning_rate": 1.7486437613019893e-05,
      "loss": 0.0116,
      "step": 3596
    },
    {
      "epoch": 4.55,
      "learning_rate": 1.747739602169982e-05,
      "loss": 0.0009,
      "step": 3597
    },
    {
      "epoch": 4.55,
      "learning_rate": 1.7468354430379746e-05,
      "loss": 0.0026,
      "step": 3598
    },
    {
      "epoch": 4.56,
      "learning_rate": 1.7459312839059674e-05,
      "loss": 0.0333,
      "step": 3599
    },
    {
      "epoch": 4.56,
      "learning_rate": 1.7450271247739602e-05,
      "loss": 0.0025,
      "step": 3600
    },
    {
      "epoch": 4.56,
      "learning_rate": 1.744122965641953e-05,
      "loss": 0.0023,
      "step": 3601
    },
    {
      "epoch": 4.56,
      "learning_rate": 1.743218806509946e-05,
      "loss": 0.1557,
      "step": 3602
    },
    {
      "epoch": 4.56,
      "learning_rate": 1.7423146473779387e-05,
      "loss": 0.0034,
      "step": 3603
    },
    {
      "epoch": 4.56,
      "learning_rate": 1.7414104882459315e-05,
      "loss": 0.2039,
      "step": 3604
    },
    {
      "epoch": 4.56,
      "learning_rate": 1.7405063291139243e-05,
      "loss": 0.003,
      "step": 3605
    },
    {
      "epoch": 4.56,
      "learning_rate": 1.7396021699819168e-05,
      "loss": 0.0324,
      "step": 3606
    },
    {
      "epoch": 4.57,
      "learning_rate": 1.7386980108499096e-05,
      "loss": 0.0017,
      "step": 3607
    },
    {
      "epoch": 4.57,
      "learning_rate": 1.7377938517179025e-05,
      "loss": 0.0211,
      "step": 3608
    },
    {
      "epoch": 4.57,
      "learning_rate": 1.7368896925858953e-05,
      "loss": 0.0772,
      "step": 3609
    },
    {
      "epoch": 4.57,
      "learning_rate": 1.735985533453888e-05,
      "loss": 0.238,
      "step": 3610
    },
    {
      "epoch": 4.57,
      "learning_rate": 1.735081374321881e-05,
      "loss": 0.0018,
      "step": 3611
    },
    {
      "epoch": 4.57,
      "learning_rate": 1.7341772151898734e-05,
      "loss": 0.0044,
      "step": 3612
    },
    {
      "epoch": 4.57,
      "learning_rate": 1.7332730560578662e-05,
      "loss": 0.0116,
      "step": 3613
    },
    {
      "epoch": 4.57,
      "learning_rate": 1.732368896925859e-05,
      "loss": 0.2074,
      "step": 3614
    },
    {
      "epoch": 4.58,
      "learning_rate": 1.7314647377938515e-05,
      "loss": 0.001,
      "step": 3615
    },
    {
      "epoch": 4.58,
      "learning_rate": 1.7305605786618444e-05,
      "loss": 0.0231,
      "step": 3616
    },
    {
      "epoch": 4.58,
      "learning_rate": 1.7296564195298372e-05,
      "loss": 0.0037,
      "step": 3617
    },
    {
      "epoch": 4.58,
      "learning_rate": 1.72875226039783e-05,
      "loss": 0.0016,
      "step": 3618
    },
    {
      "epoch": 4.58,
      "learning_rate": 1.727848101265823e-05,
      "loss": 0.0014,
      "step": 3619
    },
    {
      "epoch": 4.58,
      "learning_rate": 1.7269439421338157e-05,
      "loss": 0.0015,
      "step": 3620
    },
    {
      "epoch": 4.58,
      "learning_rate": 1.7260397830018085e-05,
      "loss": 0.0014,
      "step": 3621
    },
    {
      "epoch": 4.58,
      "learning_rate": 1.725135623869801e-05,
      "loss": 0.0008,
      "step": 3622
    },
    {
      "epoch": 4.59,
      "learning_rate": 1.7242314647377938e-05,
      "loss": 0.0036,
      "step": 3623
    },
    {
      "epoch": 4.59,
      "learning_rate": 1.7233273056057866e-05,
      "loss": 0.1593,
      "step": 3624
    },
    {
      "epoch": 4.59,
      "learning_rate": 1.7224231464737794e-05,
      "loss": 0.002,
      "step": 3625
    },
    {
      "epoch": 4.59,
      "learning_rate": 1.7215189873417723e-05,
      "loss": 0.0021,
      "step": 3626
    },
    {
      "epoch": 4.59,
      "learning_rate": 1.720614828209765e-05,
      "loss": 0.0012,
      "step": 3627
    },
    {
      "epoch": 4.59,
      "learning_rate": 1.719710669077758e-05,
      "loss": 0.001,
      "step": 3628
    },
    {
      "epoch": 4.59,
      "learning_rate": 1.7188065099457507e-05,
      "loss": 0.0014,
      "step": 3629
    },
    {
      "epoch": 4.59,
      "learning_rate": 1.7179023508137432e-05,
      "loss": 0.143,
      "step": 3630
    },
    {
      "epoch": 4.6,
      "learning_rate": 1.716998191681736e-05,
      "loss": 0.0035,
      "step": 3631
    },
    {
      "epoch": 4.6,
      "learning_rate": 1.716094032549729e-05,
      "loss": 0.0005,
      "step": 3632
    },
    {
      "epoch": 4.6,
      "learning_rate": 1.7151898734177217e-05,
      "loss": 0.0321,
      "step": 3633
    },
    {
      "epoch": 4.6,
      "learning_rate": 1.7142857142857145e-05,
      "loss": 0.0008,
      "step": 3634
    },
    {
      "epoch": 4.6,
      "learning_rate": 1.7133815551537073e-05,
      "loss": 0.0015,
      "step": 3635
    },
    {
      "epoch": 4.6,
      "learning_rate": 1.7124773960217e-05,
      "loss": 0.0071,
      "step": 3636
    },
    {
      "epoch": 4.6,
      "learning_rate": 1.7115732368896927e-05,
      "loss": 0.0095,
      "step": 3637
    },
    {
      "epoch": 4.6,
      "learning_rate": 1.7106690777576855e-05,
      "loss": 0.0011,
      "step": 3638
    },
    {
      "epoch": 4.61,
      "learning_rate": 1.709764918625678e-05,
      "loss": 0.001,
      "step": 3639
    },
    {
      "epoch": 4.61,
      "learning_rate": 1.7088607594936708e-05,
      "loss": 0.0369,
      "step": 3640
    },
    {
      "epoch": 4.61,
      "learning_rate": 1.7079566003616636e-05,
      "loss": 0.0018,
      "step": 3641
    },
    {
      "epoch": 4.61,
      "learning_rate": 1.7070524412296564e-05,
      "loss": 0.0011,
      "step": 3642
    },
    {
      "epoch": 4.61,
      "learning_rate": 1.7061482820976493e-05,
      "loss": 0.0006,
      "step": 3643
    },
    {
      "epoch": 4.61,
      "learning_rate": 1.705244122965642e-05,
      "loss": 0.0012,
      "step": 3644
    },
    {
      "epoch": 4.61,
      "learning_rate": 1.704339963833635e-05,
      "loss": 0.0006,
      "step": 3645
    },
    {
      "epoch": 4.61,
      "learning_rate": 1.7034358047016274e-05,
      "loss": 0.1701,
      "step": 3646
    },
    {
      "epoch": 4.62,
      "learning_rate": 1.7025316455696202e-05,
      "loss": 0.0009,
      "step": 3647
    },
    {
      "epoch": 4.62,
      "learning_rate": 1.701627486437613e-05,
      "loss": 0.1018,
      "step": 3648
    },
    {
      "epoch": 4.62,
      "learning_rate": 1.700723327305606e-05,
      "loss": 0.2343,
      "step": 3649
    },
    {
      "epoch": 4.62,
      "learning_rate": 1.6998191681735987e-05,
      "loss": 0.0011,
      "step": 3650
    },
    {
      "epoch": 4.62,
      "learning_rate": 1.6989150090415915e-05,
      "loss": 0.0009,
      "step": 3651
    },
    {
      "epoch": 4.62,
      "learning_rate": 1.6980108499095843e-05,
      "loss": 0.0008,
      "step": 3652
    },
    {
      "epoch": 4.62,
      "learning_rate": 1.697106690777577e-05,
      "loss": 0.0043,
      "step": 3653
    },
    {
      "epoch": 4.62,
      "learning_rate": 1.6962025316455696e-05,
      "loss": 0.0035,
      "step": 3654
    },
    {
      "epoch": 4.63,
      "learning_rate": 1.6952983725135625e-05,
      "loss": 0.0979,
      "step": 3655
    },
    {
      "epoch": 4.63,
      "learning_rate": 1.6943942133815553e-05,
      "loss": 0.0496,
      "step": 3656
    },
    {
      "epoch": 4.63,
      "learning_rate": 1.693490054249548e-05,
      "loss": 0.1233,
      "step": 3657
    },
    {
      "epoch": 4.63,
      "learning_rate": 1.692585895117541e-05,
      "loss": 0.0025,
      "step": 3658
    },
    {
      "epoch": 4.63,
      "learning_rate": 1.6916817359855334e-05,
      "loss": 0.0008,
      "step": 3659
    },
    {
      "epoch": 4.63,
      "learning_rate": 1.6907775768535262e-05,
      "loss": 0.0039,
      "step": 3660
    },
    {
      "epoch": 4.63,
      "learning_rate": 1.689873417721519e-05,
      "loss": 0.0138,
      "step": 3661
    },
    {
      "epoch": 4.64,
      "learning_rate": 1.6889692585895116e-05,
      "loss": 0.0018,
      "step": 3662
    },
    {
      "epoch": 4.64,
      "learning_rate": 1.6880650994575044e-05,
      "loss": 0.1664,
      "step": 3663
    },
    {
      "epoch": 4.64,
      "learning_rate": 1.6871609403254972e-05,
      "loss": 0.0119,
      "step": 3664
    },
    {
      "epoch": 4.64,
      "learning_rate": 1.68625678119349e-05,
      "loss": 0.0074,
      "step": 3665
    },
    {
      "epoch": 4.64,
      "learning_rate": 1.685352622061483e-05,
      "loss": 0.001,
      "step": 3666
    },
    {
      "epoch": 4.64,
      "learning_rate": 1.6844484629294757e-05,
      "loss": 0.0007,
      "step": 3667
    },
    {
      "epoch": 4.64,
      "learning_rate": 1.6835443037974685e-05,
      "loss": 0.0011,
      "step": 3668
    },
    {
      "epoch": 4.64,
      "learning_rate": 1.6826401446654613e-05,
      "loss": 0.0022,
      "step": 3669
    },
    {
      "epoch": 4.65,
      "learning_rate": 1.6817359855334538e-05,
      "loss": 0.001,
      "step": 3670
    },
    {
      "epoch": 4.65,
      "learning_rate": 1.6808318264014466e-05,
      "loss": 0.0011,
      "step": 3671
    },
    {
      "epoch": 4.65,
      "learning_rate": 1.6799276672694395e-05,
      "loss": 0.0006,
      "step": 3672
    },
    {
      "epoch": 4.65,
      "learning_rate": 1.6790235081374323e-05,
      "loss": 0.012,
      "step": 3673
    },
    {
      "epoch": 4.65,
      "learning_rate": 1.678119349005425e-05,
      "loss": 0.0027,
      "step": 3674
    },
    {
      "epoch": 4.65,
      "learning_rate": 1.677215189873418e-05,
      "loss": 0.0013,
      "step": 3675
    },
    {
      "epoch": 4.65,
      "learning_rate": 1.6763110307414108e-05,
      "loss": 0.0023,
      "step": 3676
    },
    {
      "epoch": 4.65,
      "learning_rate": 1.6754068716094036e-05,
      "loss": 0.0015,
      "step": 3677
    },
    {
      "epoch": 4.66,
      "learning_rate": 1.674502712477396e-05,
      "loss": 0.0013,
      "step": 3678
    },
    {
      "epoch": 4.66,
      "learning_rate": 1.673598553345389e-05,
      "loss": 0.1692,
      "step": 3679
    },
    {
      "epoch": 4.66,
      "learning_rate": 1.6726943942133817e-05,
      "loss": 0.002,
      "step": 3680
    },
    {
      "epoch": 4.66,
      "learning_rate": 1.6717902350813745e-05,
      "loss": 0.0009,
      "step": 3681
    },
    {
      "epoch": 4.66,
      "learning_rate": 1.6708860759493674e-05,
      "loss": 0.0032,
      "step": 3682
    },
    {
      "epoch": 4.66,
      "learning_rate": 1.66998191681736e-05,
      "loss": 0.0036,
      "step": 3683
    },
    {
      "epoch": 4.66,
      "learning_rate": 1.6690777576853527e-05,
      "loss": 0.0013,
      "step": 3684
    },
    {
      "epoch": 4.66,
      "learning_rate": 1.6681735985533455e-05,
      "loss": 0.0014,
      "step": 3685
    },
    {
      "epoch": 4.67,
      "learning_rate": 1.667269439421338e-05,
      "loss": 0.132,
      "step": 3686
    },
    {
      "epoch": 4.67,
      "learning_rate": 1.6663652802893308e-05,
      "loss": 0.0052,
      "step": 3687
    },
    {
      "epoch": 4.67,
      "learning_rate": 1.6654611211573236e-05,
      "loss": 0.0024,
      "step": 3688
    },
    {
      "epoch": 4.67,
      "learning_rate": 1.6645569620253164e-05,
      "loss": 0.002,
      "step": 3689
    },
    {
      "epoch": 4.67,
      "learning_rate": 1.6636528028933093e-05,
      "loss": 0.0009,
      "step": 3690
    },
    {
      "epoch": 4.67,
      "learning_rate": 1.662748643761302e-05,
      "loss": 0.0004,
      "step": 3691
    },
    {
      "epoch": 4.67,
      "learning_rate": 1.661844484629295e-05,
      "loss": 0.0017,
      "step": 3692
    },
    {
      "epoch": 4.67,
      "learning_rate": 1.6609403254972877e-05,
      "loss": 0.0031,
      "step": 3693
    },
    {
      "epoch": 4.68,
      "learning_rate": 1.6600361663652802e-05,
      "loss": 0.2374,
      "step": 3694
    },
    {
      "epoch": 4.68,
      "learning_rate": 1.659132007233273e-05,
      "loss": 0.0006,
      "step": 3695
    },
    {
      "epoch": 4.68,
      "learning_rate": 1.658227848101266e-05,
      "loss": 0.1486,
      "step": 3696
    },
    {
      "epoch": 4.68,
      "learning_rate": 1.6573236889692587e-05,
      "loss": 0.0012,
      "step": 3697
    },
    {
      "epoch": 4.68,
      "learning_rate": 1.6564195298372515e-05,
      "loss": 0.0234,
      "step": 3698
    },
    {
      "epoch": 4.68,
      "learning_rate": 1.6555153707052443e-05,
      "loss": 0.003,
      "step": 3699
    },
    {
      "epoch": 4.68,
      "learning_rate": 1.654611211573237e-05,
      "loss": 0.0012,
      "step": 3700
    },
    {
      "epoch": 4.68,
      "learning_rate": 1.65370705244123e-05,
      "loss": 0.0057,
      "step": 3701
    },
    {
      "epoch": 4.69,
      "learning_rate": 1.6528028933092225e-05,
      "loss": 0.0017,
      "step": 3702
    },
    {
      "epoch": 4.69,
      "learning_rate": 1.6518987341772153e-05,
      "loss": 0.0071,
      "step": 3703
    },
    {
      "epoch": 4.69,
      "learning_rate": 1.650994575045208e-05,
      "loss": 0.0012,
      "step": 3704
    },
    {
      "epoch": 4.69,
      "learning_rate": 1.650090415913201e-05,
      "loss": 0.0263,
      "step": 3705
    },
    {
      "epoch": 4.69,
      "learning_rate": 1.6491862567811934e-05,
      "loss": 0.0014,
      "step": 3706
    },
    {
      "epoch": 4.69,
      "learning_rate": 1.6482820976491863e-05,
      "loss": 0.001,
      "step": 3707
    },
    {
      "epoch": 4.69,
      "learning_rate": 1.647377938517179e-05,
      "loss": 0.0014,
      "step": 3708
    },
    {
      "epoch": 4.69,
      "learning_rate": 1.646473779385172e-05,
      "loss": 0.2267,
      "step": 3709
    },
    {
      "epoch": 4.7,
      "learning_rate": 1.6455696202531644e-05,
      "loss": 0.2114,
      "step": 3710
    },
    {
      "epoch": 4.7,
      "learning_rate": 1.6446654611211572e-05,
      "loss": 0.0007,
      "step": 3711
    },
    {
      "epoch": 4.7,
      "learning_rate": 1.64376130198915e-05,
      "loss": 0.0009,
      "step": 3712
    },
    {
      "epoch": 4.7,
      "learning_rate": 1.642857142857143e-05,
      "loss": 0.0017,
      "step": 3713
    },
    {
      "epoch": 4.7,
      "learning_rate": 1.6419529837251357e-05,
      "loss": 0.0022,
      "step": 3714
    },
    {
      "epoch": 4.7,
      "learning_rate": 1.6410488245931285e-05,
      "loss": 0.0087,
      "step": 3715
    },
    {
      "epoch": 4.7,
      "learning_rate": 1.6401446654611213e-05,
      "loss": 0.0668,
      "step": 3716
    },
    {
      "epoch": 4.7,
      "learning_rate": 1.639240506329114e-05,
      "loss": 0.197,
      "step": 3717
    },
    {
      "epoch": 4.71,
      "learning_rate": 1.6383363471971066e-05,
      "loss": 0.0091,
      "step": 3718
    },
    {
      "epoch": 4.71,
      "learning_rate": 1.6374321880650995e-05,
      "loss": 0.0031,
      "step": 3719
    },
    {
      "epoch": 4.71,
      "learning_rate": 1.6365280289330923e-05,
      "loss": 0.0216,
      "step": 3720
    },
    {
      "epoch": 4.71,
      "learning_rate": 1.635623869801085e-05,
      "loss": 0.001,
      "step": 3721
    },
    {
      "epoch": 4.71,
      "learning_rate": 1.634719710669078e-05,
      "loss": 0.0009,
      "step": 3722
    },
    {
      "epoch": 4.71,
      "learning_rate": 1.6338155515370708e-05,
      "loss": 0.0008,
      "step": 3723
    },
    {
      "epoch": 4.71,
      "learning_rate": 1.6329113924050636e-05,
      "loss": 0.0007,
      "step": 3724
    },
    {
      "epoch": 4.71,
      "learning_rate": 1.6320072332730564e-05,
      "loss": 0.001,
      "step": 3725
    },
    {
      "epoch": 4.72,
      "learning_rate": 1.631103074141049e-05,
      "loss": 0.0026,
      "step": 3726
    },
    {
      "epoch": 4.72,
      "learning_rate": 1.6301989150090417e-05,
      "loss": 0.0009,
      "step": 3727
    },
    {
      "epoch": 4.72,
      "learning_rate": 1.6292947558770345e-05,
      "loss": 0.0009,
      "step": 3728
    },
    {
      "epoch": 4.72,
      "learning_rate": 1.6283905967450274e-05,
      "loss": 0.0016,
      "step": 3729
    },
    {
      "epoch": 4.72,
      "learning_rate": 1.62748643761302e-05,
      "loss": 0.0006,
      "step": 3730
    },
    {
      "epoch": 4.72,
      "learning_rate": 1.6265822784810127e-05,
      "loss": 0.0488,
      "step": 3731
    },
    {
      "epoch": 4.72,
      "learning_rate": 1.6256781193490055e-05,
      "loss": 0.0015,
      "step": 3732
    },
    {
      "epoch": 4.72,
      "learning_rate": 1.6247739602169983e-05,
      "loss": 0.0014,
      "step": 3733
    },
    {
      "epoch": 4.73,
      "learning_rate": 1.6238698010849908e-05,
      "loss": 0.1856,
      "step": 3734
    },
    {
      "epoch": 4.73,
      "learning_rate": 1.6229656419529836e-05,
      "loss": 0.0194,
      "step": 3735
    },
    {
      "epoch": 4.73,
      "learning_rate": 1.6220614828209765e-05,
      "loss": 0.0024,
      "step": 3736
    },
    {
      "epoch": 4.73,
      "learning_rate": 1.6211573236889693e-05,
      "loss": 0.0014,
      "step": 3737
    },
    {
      "epoch": 4.73,
      "learning_rate": 1.620253164556962e-05,
      "loss": 0.002,
      "step": 3738
    },
    {
      "epoch": 4.73,
      "learning_rate": 1.619349005424955e-05,
      "loss": 0.0005,
      "step": 3739
    },
    {
      "epoch": 4.73,
      "learning_rate": 1.6184448462929477e-05,
      "loss": 0.0007,
      "step": 3740
    },
    {
      "epoch": 4.73,
      "learning_rate": 1.6175406871609406e-05,
      "loss": 0.0018,
      "step": 3741
    },
    {
      "epoch": 4.74,
      "learning_rate": 1.616636528028933e-05,
      "loss": 0.0039,
      "step": 3742
    },
    {
      "epoch": 4.74,
      "learning_rate": 1.615732368896926e-05,
      "loss": 0.0036,
      "step": 3743
    },
    {
      "epoch": 4.74,
      "learning_rate": 1.6148282097649187e-05,
      "loss": 0.0014,
      "step": 3744
    },
    {
      "epoch": 4.74,
      "learning_rate": 1.6139240506329115e-05,
      "loss": 0.0007,
      "step": 3745
    },
    {
      "epoch": 4.74,
      "learning_rate": 1.6130198915009043e-05,
      "loss": 0.0012,
      "step": 3746
    },
    {
      "epoch": 4.74,
      "learning_rate": 1.6121157323688972e-05,
      "loss": 0.0068,
      "step": 3747
    },
    {
      "epoch": 4.74,
      "learning_rate": 1.61121157323689e-05,
      "loss": 0.0032,
      "step": 3748
    },
    {
      "epoch": 4.75,
      "learning_rate": 1.6103074141048828e-05,
      "loss": 0.0004,
      "step": 3749
    },
    {
      "epoch": 4.75,
      "learning_rate": 1.6094032549728753e-05,
      "loss": 0.0009,
      "step": 3750
    },
    {
      "epoch": 4.75,
      "eval_accuracy": 0.9046546546546547,
      "eval_loss": 0.4625249207019806,
      "eval_runtime": 45.0773,
      "eval_samples_per_second": 29.549,
      "eval_steps_per_second": 3.705,
      "step": 3750
    },
    {
      "epoch": 4.75,
      "learning_rate": 1.608499095840868e-05,
      "loss": 0.0237,
      "step": 3751
    },
    {
      "epoch": 4.75,
      "learning_rate": 1.607594936708861e-05,
      "loss": 0.0004,
      "step": 3752
    },
    {
      "epoch": 4.75,
      "learning_rate": 1.6066907775768534e-05,
      "loss": 0.0028,
      "step": 3753
    },
    {
      "epoch": 4.75,
      "learning_rate": 1.6057866184448463e-05,
      "loss": 0.0008,
      "step": 3754
    },
    {
      "epoch": 4.75,
      "learning_rate": 1.604882459312839e-05,
      "loss": 0.0007,
      "step": 3755
    },
    {
      "epoch": 4.75,
      "learning_rate": 1.603978300180832e-05,
      "loss": 0.0007,
      "step": 3756
    },
    {
      "epoch": 4.76,
      "learning_rate": 1.6030741410488247e-05,
      "loss": 0.0697,
      "step": 3757
    },
    {
      "epoch": 4.76,
      "learning_rate": 1.6021699819168172e-05,
      "loss": 0.0006,
      "step": 3758
    },
    {
      "epoch": 4.76,
      "learning_rate": 1.60126582278481e-05,
      "loss": 0.0024,
      "step": 3759
    },
    {
      "epoch": 4.76,
      "learning_rate": 1.600361663652803e-05,
      "loss": 0.0011,
      "step": 3760
    },
    {
      "epoch": 4.76,
      "learning_rate": 1.5994575045207957e-05,
      "loss": 0.0006,
      "step": 3761
    },
    {
      "epoch": 4.76,
      "learning_rate": 1.5985533453887885e-05,
      "loss": 0.0011,
      "step": 3762
    },
    {
      "epoch": 4.76,
      "learning_rate": 1.5976491862567813e-05,
      "loss": 0.0011,
      "step": 3763
    },
    {
      "epoch": 4.76,
      "learning_rate": 1.596745027124774e-05,
      "loss": 0.0008,
      "step": 3764
    },
    {
      "epoch": 4.77,
      "learning_rate": 1.595840867992767e-05,
      "loss": 0.1056,
      "step": 3765
    },
    {
      "epoch": 4.77,
      "learning_rate": 1.5949367088607595e-05,
      "loss": 0.0008,
      "step": 3766
    },
    {
      "epoch": 4.77,
      "learning_rate": 1.5940325497287523e-05,
      "loss": 0.1291,
      "step": 3767
    },
    {
      "epoch": 4.77,
      "learning_rate": 1.593128390596745e-05,
      "loss": 0.0136,
      "step": 3768
    },
    {
      "epoch": 4.77,
      "learning_rate": 1.592224231464738e-05,
      "loss": 0.0005,
      "step": 3769
    },
    {
      "epoch": 4.77,
      "learning_rate": 1.5913200723327308e-05,
      "loss": 0.0225,
      "step": 3770
    },
    {
      "epoch": 4.77,
      "learning_rate": 1.5904159132007236e-05,
      "loss": 0.0006,
      "step": 3771
    },
    {
      "epoch": 4.77,
      "learning_rate": 1.5895117540687164e-05,
      "loss": 0.0012,
      "step": 3772
    },
    {
      "epoch": 4.78,
      "learning_rate": 1.588607594936709e-05,
      "loss": 0.001,
      "step": 3773
    },
    {
      "epoch": 4.78,
      "learning_rate": 1.5877034358047017e-05,
      "loss": 0.0008,
      "step": 3774
    },
    {
      "epoch": 4.78,
      "learning_rate": 1.5867992766726945e-05,
      "loss": 0.0578,
      "step": 3775
    },
    {
      "epoch": 4.78,
      "learning_rate": 1.5858951175406874e-05,
      "loss": 0.0029,
      "step": 3776
    },
    {
      "epoch": 4.78,
      "learning_rate": 1.58499095840868e-05,
      "loss": 0.0011,
      "step": 3777
    },
    {
      "epoch": 4.78,
      "learning_rate": 1.5840867992766727e-05,
      "loss": 0.0035,
      "step": 3778
    },
    {
      "epoch": 4.78,
      "learning_rate": 1.5831826401446655e-05,
      "loss": 0.0008,
      "step": 3779
    },
    {
      "epoch": 4.78,
      "learning_rate": 1.5822784810126583e-05,
      "loss": 0.001,
      "step": 3780
    },
    {
      "epoch": 4.79,
      "learning_rate": 1.5813743218806508e-05,
      "loss": 0.0008,
      "step": 3781
    },
    {
      "epoch": 4.79,
      "learning_rate": 1.5804701627486436e-05,
      "loss": 0.0023,
      "step": 3782
    },
    {
      "epoch": 4.79,
      "learning_rate": 1.5795660036166365e-05,
      "loss": 0.0009,
      "step": 3783
    },
    {
      "epoch": 4.79,
      "learning_rate": 1.5786618444846293e-05,
      "loss": 0.2112,
      "step": 3784
    },
    {
      "epoch": 4.79,
      "learning_rate": 1.577757685352622e-05,
      "loss": 0.1084,
      "step": 3785
    },
    {
      "epoch": 4.79,
      "learning_rate": 1.576853526220615e-05,
      "loss": 0.036,
      "step": 3786
    },
    {
      "epoch": 4.79,
      "learning_rate": 1.5759493670886078e-05,
      "loss": 0.0011,
      "step": 3787
    },
    {
      "epoch": 4.79,
      "learning_rate": 1.5750452079566006e-05,
      "loss": 0.1541,
      "step": 3788
    },
    {
      "epoch": 4.8,
      "learning_rate": 1.574141048824593e-05,
      "loss": 0.0007,
      "step": 3789
    },
    {
      "epoch": 4.8,
      "learning_rate": 1.573236889692586e-05,
      "loss": 0.0008,
      "step": 3790
    },
    {
      "epoch": 4.8,
      "learning_rate": 1.5723327305605787e-05,
      "loss": 0.0005,
      "step": 3791
    },
    {
      "epoch": 4.8,
      "learning_rate": 1.5714285714285715e-05,
      "loss": 0.0007,
      "step": 3792
    },
    {
      "epoch": 4.8,
      "learning_rate": 1.5705244122965644e-05,
      "loss": 0.0007,
      "step": 3793
    },
    {
      "epoch": 4.8,
      "learning_rate": 1.5696202531645572e-05,
      "loss": 0.0006,
      "step": 3794
    },
    {
      "epoch": 4.8,
      "learning_rate": 1.56871609403255e-05,
      "loss": 0.0006,
      "step": 3795
    },
    {
      "epoch": 4.8,
      "learning_rate": 1.5678119349005428e-05,
      "loss": 0.0015,
      "step": 3796
    },
    {
      "epoch": 4.81,
      "learning_rate": 1.5669077757685353e-05,
      "loss": 0.0007,
      "step": 3797
    },
    {
      "epoch": 4.81,
      "learning_rate": 1.566003616636528e-05,
      "loss": 0.0022,
      "step": 3798
    },
    {
      "epoch": 4.81,
      "learning_rate": 1.565099457504521e-05,
      "loss": 0.0298,
      "step": 3799
    },
    {
      "epoch": 4.81,
      "learning_rate": 1.5641952983725134e-05,
      "loss": 0.0018,
      "step": 3800
    },
    {
      "epoch": 4.81,
      "learning_rate": 1.5632911392405063e-05,
      "loss": 0.3153,
      "step": 3801
    },
    {
      "epoch": 4.81,
      "learning_rate": 1.562386980108499e-05,
      "loss": 0.0095,
      "step": 3802
    },
    {
      "epoch": 4.81,
      "learning_rate": 1.561482820976492e-05,
      "loss": 0.3266,
      "step": 3803
    },
    {
      "epoch": 4.81,
      "learning_rate": 1.5605786618444847e-05,
      "loss": 0.0011,
      "step": 3804
    },
    {
      "epoch": 4.82,
      "learning_rate": 1.5596745027124772e-05,
      "loss": 0.0683,
      "step": 3805
    },
    {
      "epoch": 4.82,
      "learning_rate": 1.55877034358047e-05,
      "loss": 0.0054,
      "step": 3806
    },
    {
      "epoch": 4.82,
      "learning_rate": 1.557866184448463e-05,
      "loss": 0.0016,
      "step": 3807
    },
    {
      "epoch": 4.82,
      "learning_rate": 1.5569620253164557e-05,
      "loss": 0.0191,
      "step": 3808
    },
    {
      "epoch": 4.82,
      "learning_rate": 1.5560578661844485e-05,
      "loss": 0.0015,
      "step": 3809
    },
    {
      "epoch": 4.82,
      "learning_rate": 1.5551537070524413e-05,
      "loss": 0.0019,
      "step": 3810
    },
    {
      "epoch": 4.82,
      "learning_rate": 1.554249547920434e-05,
      "loss": 0.1172,
      "step": 3811
    },
    {
      "epoch": 4.82,
      "learning_rate": 1.553345388788427e-05,
      "loss": 0.1424,
      "step": 3812
    },
    {
      "epoch": 4.83,
      "learning_rate": 1.5524412296564195e-05,
      "loss": 0.2257,
      "step": 3813
    },
    {
      "epoch": 4.83,
      "learning_rate": 1.5515370705244123e-05,
      "loss": 0.1585,
      "step": 3814
    },
    {
      "epoch": 4.83,
      "learning_rate": 1.550632911392405e-05,
      "loss": 0.0008,
      "step": 3815
    },
    {
      "epoch": 4.83,
      "learning_rate": 1.549728752260398e-05,
      "loss": 0.0059,
      "step": 3816
    },
    {
      "epoch": 4.83,
      "learning_rate": 1.5488245931283908e-05,
      "loss": 0.0009,
      "step": 3817
    },
    {
      "epoch": 4.83,
      "learning_rate": 1.5479204339963836e-05,
      "loss": 0.1663,
      "step": 3818
    },
    {
      "epoch": 4.83,
      "learning_rate": 1.5470162748643764e-05,
      "loss": 0.0018,
      "step": 3819
    },
    {
      "epoch": 4.83,
      "learning_rate": 1.5461121157323692e-05,
      "loss": 0.0036,
      "step": 3820
    },
    {
      "epoch": 4.84,
      "learning_rate": 1.5452079566003617e-05,
      "loss": 0.0004,
      "step": 3821
    },
    {
      "epoch": 4.84,
      "learning_rate": 1.5443037974683546e-05,
      "loss": 0.0023,
      "step": 3822
    },
    {
      "epoch": 4.84,
      "learning_rate": 1.5433996383363474e-05,
      "loss": 0.002,
      "step": 3823
    },
    {
      "epoch": 4.84,
      "learning_rate": 1.54249547920434e-05,
      "loss": 0.0021,
      "step": 3824
    },
    {
      "epoch": 4.84,
      "learning_rate": 1.5415913200723327e-05,
      "loss": 0.0013,
      "step": 3825
    },
    {
      "epoch": 4.84,
      "learning_rate": 1.5406871609403255e-05,
      "loss": 0.0888,
      "step": 3826
    },
    {
      "epoch": 4.84,
      "learning_rate": 1.5397830018083183e-05,
      "loss": 0.0005,
      "step": 3827
    },
    {
      "epoch": 4.85,
      "learning_rate": 1.538878842676311e-05,
      "loss": 0.0018,
      "step": 3828
    },
    {
      "epoch": 4.85,
      "learning_rate": 1.5379746835443036e-05,
      "loss": 0.0405,
      "step": 3829
    },
    {
      "epoch": 4.85,
      "learning_rate": 1.5370705244122965e-05,
      "loss": 0.0011,
      "step": 3830
    },
    {
      "epoch": 4.85,
      "learning_rate": 1.5361663652802893e-05,
      "loss": 0.0006,
      "step": 3831
    },
    {
      "epoch": 4.85,
      "learning_rate": 1.535262206148282e-05,
      "loss": 0.1112,
      "step": 3832
    },
    {
      "epoch": 4.85,
      "learning_rate": 1.534358047016275e-05,
      "loss": 0.0085,
      "step": 3833
    },
    {
      "epoch": 4.85,
      "learning_rate": 1.5334538878842678e-05,
      "loss": 0.0237,
      "step": 3834
    },
    {
      "epoch": 4.85,
      "learning_rate": 1.5325497287522606e-05,
      "loss": 0.0528,
      "step": 3835
    },
    {
      "epoch": 4.86,
      "learning_rate": 1.5316455696202534e-05,
      "loss": 0.0039,
      "step": 3836
    },
    {
      "epoch": 4.86,
      "learning_rate": 1.530741410488246e-05,
      "loss": 0.0046,
      "step": 3837
    },
    {
      "epoch": 4.86,
      "learning_rate": 1.5298372513562387e-05,
      "loss": 0.0036,
      "step": 3838
    },
    {
      "epoch": 4.86,
      "learning_rate": 1.5289330922242315e-05,
      "loss": 0.1322,
      "step": 3839
    },
    {
      "epoch": 4.86,
      "learning_rate": 1.5280289330922244e-05,
      "loss": 0.0008,
      "step": 3840
    },
    {
      "epoch": 4.86,
      "learning_rate": 1.5271247739602172e-05,
      "loss": 0.001,
      "step": 3841
    },
    {
      "epoch": 4.86,
      "learning_rate": 1.52622061482821e-05,
      "loss": 0.0013,
      "step": 3842
    },
    {
      "epoch": 4.86,
      "learning_rate": 1.5253164556962027e-05,
      "loss": 0.1668,
      "step": 3843
    },
    {
      "epoch": 4.87,
      "learning_rate": 1.5244122965641955e-05,
      "loss": 0.1349,
      "step": 3844
    },
    {
      "epoch": 4.87,
      "learning_rate": 1.523508137432188e-05,
      "loss": 0.001,
      "step": 3845
    },
    {
      "epoch": 4.87,
      "learning_rate": 1.5226039783001808e-05,
      "loss": 0.04,
      "step": 3846
    },
    {
      "epoch": 4.87,
      "learning_rate": 1.5216998191681736e-05,
      "loss": 0.0094,
      "step": 3847
    },
    {
      "epoch": 4.87,
      "learning_rate": 1.5207956600361664e-05,
      "loss": 0.0097,
      "step": 3848
    },
    {
      "epoch": 4.87,
      "learning_rate": 1.5198915009041593e-05,
      "loss": 0.023,
      "step": 3849
    },
    {
      "epoch": 4.87,
      "learning_rate": 1.5189873417721521e-05,
      "loss": 0.0021,
      "step": 3850
    },
    {
      "epoch": 4.87,
      "learning_rate": 1.5180831826401447e-05,
      "loss": 0.0278,
      "step": 3851
    },
    {
      "epoch": 4.88,
      "learning_rate": 1.5171790235081376e-05,
      "loss": 0.0016,
      "step": 3852
    },
    {
      "epoch": 4.88,
      "learning_rate": 1.5162748643761302e-05,
      "loss": 0.0024,
      "step": 3853
    },
    {
      "epoch": 4.88,
      "learning_rate": 1.5153707052441229e-05,
      "loss": 0.007,
      "step": 3854
    },
    {
      "epoch": 4.88,
      "learning_rate": 1.5144665461121157e-05,
      "loss": 0.001,
      "step": 3855
    },
    {
      "epoch": 4.88,
      "learning_rate": 1.5135623869801085e-05,
      "loss": 0.0004,
      "step": 3856
    },
    {
      "epoch": 4.88,
      "learning_rate": 1.5126582278481014e-05,
      "loss": 0.0008,
      "step": 3857
    },
    {
      "epoch": 4.88,
      "learning_rate": 1.5117540687160942e-05,
      "loss": 0.0077,
      "step": 3858
    },
    {
      "epoch": 4.88,
      "learning_rate": 1.510849909584087e-05,
      "loss": 0.0021,
      "step": 3859
    },
    {
      "epoch": 4.89,
      "learning_rate": 1.5099457504520798e-05,
      "loss": 0.0023,
      "step": 3860
    },
    {
      "epoch": 4.89,
      "learning_rate": 1.5090415913200723e-05,
      "loss": 0.0011,
      "step": 3861
    },
    {
      "epoch": 4.89,
      "learning_rate": 1.5081374321880651e-05,
      "loss": 0.0012,
      "step": 3862
    },
    {
      "epoch": 4.89,
      "learning_rate": 1.507233273056058e-05,
      "loss": 0.0013,
      "step": 3863
    },
    {
      "epoch": 4.89,
      "learning_rate": 1.5063291139240506e-05,
      "loss": 0.0016,
      "step": 3864
    },
    {
      "epoch": 4.89,
      "learning_rate": 1.5054249547920434e-05,
      "loss": 0.0209,
      "step": 3865
    },
    {
      "epoch": 4.89,
      "learning_rate": 1.5045207956600363e-05,
      "loss": 0.0011,
      "step": 3866
    },
    {
      "epoch": 4.89,
      "learning_rate": 1.503616636528029e-05,
      "loss": 0.0011,
      "step": 3867
    },
    {
      "epoch": 4.9,
      "learning_rate": 1.5027124773960219e-05,
      "loss": 0.0007,
      "step": 3868
    },
    {
      "epoch": 4.9,
      "learning_rate": 1.5018083182640144e-05,
      "loss": 0.0031,
      "step": 3869
    },
    {
      "epoch": 4.9,
      "learning_rate": 1.5009041591320072e-05,
      "loss": 0.0011,
      "step": 3870
    },
    {
      "epoch": 4.9,
      "learning_rate": 1.5e-05,
      "loss": 0.0005,
      "step": 3871
    },
    {
      "epoch": 4.9,
      "learning_rate": 1.4990958408679929e-05,
      "loss": 0.0011,
      "step": 3872
    },
    {
      "epoch": 4.9,
      "learning_rate": 1.4981916817359857e-05,
      "loss": 0.0007,
      "step": 3873
    },
    {
      "epoch": 4.9,
      "learning_rate": 1.4972875226039783e-05,
      "loss": 0.0003,
      "step": 3874
    },
    {
      "epoch": 4.9,
      "learning_rate": 1.4963833634719712e-05,
      "loss": 0.0181,
      "step": 3875
    },
    {
      "epoch": 4.91,
      "learning_rate": 1.495479204339964e-05,
      "loss": 0.0008,
      "step": 3876
    },
    {
      "epoch": 4.91,
      "learning_rate": 1.4945750452079565e-05,
      "loss": 0.0862,
      "step": 3877
    },
    {
      "epoch": 4.91,
      "learning_rate": 1.4936708860759493e-05,
      "loss": 0.0598,
      "step": 3878
    },
    {
      "epoch": 4.91,
      "learning_rate": 1.4927667269439421e-05,
      "loss": 0.0019,
      "step": 3879
    },
    {
      "epoch": 4.91,
      "learning_rate": 1.491862567811935e-05,
      "loss": 0.0019,
      "step": 3880
    },
    {
      "epoch": 4.91,
      "learning_rate": 1.4909584086799278e-05,
      "loss": 0.0008,
      "step": 3881
    },
    {
      "epoch": 4.91,
      "learning_rate": 1.4900542495479206e-05,
      "loss": 0.1054,
      "step": 3882
    },
    {
      "epoch": 4.91,
      "learning_rate": 1.4891500904159134e-05,
      "loss": 0.1852,
      "step": 3883
    },
    {
      "epoch": 4.92,
      "learning_rate": 1.4882459312839062e-05,
      "loss": 0.1289,
      "step": 3884
    },
    {
      "epoch": 4.92,
      "learning_rate": 1.4873417721518987e-05,
      "loss": 0.0049,
      "step": 3885
    },
    {
      "epoch": 4.92,
      "learning_rate": 1.4864376130198915e-05,
      "loss": 0.213,
      "step": 3886
    },
    {
      "epoch": 4.92,
      "learning_rate": 1.4855334538878842e-05,
      "loss": 0.0181,
      "step": 3887
    },
    {
      "epoch": 4.92,
      "learning_rate": 1.484629294755877e-05,
      "loss": 0.0007,
      "step": 3888
    },
    {
      "epoch": 4.92,
      "learning_rate": 1.4837251356238698e-05,
      "loss": 0.0006,
      "step": 3889
    },
    {
      "epoch": 4.92,
      "learning_rate": 1.4828209764918627e-05,
      "loss": 0.0006,
      "step": 3890
    },
    {
      "epoch": 4.92,
      "learning_rate": 1.4819168173598555e-05,
      "loss": 0.0018,
      "step": 3891
    },
    {
      "epoch": 4.93,
      "learning_rate": 1.4810126582278483e-05,
      "loss": 0.0016,
      "step": 3892
    },
    {
      "epoch": 4.93,
      "learning_rate": 1.4801084990958408e-05,
      "loss": 0.0018,
      "step": 3893
    },
    {
      "epoch": 4.93,
      "learning_rate": 1.4792043399638336e-05,
      "loss": 0.0005,
      "step": 3894
    },
    {
      "epoch": 4.93,
      "learning_rate": 1.4783001808318265e-05,
      "loss": 0.0045,
      "step": 3895
    },
    {
      "epoch": 4.93,
      "learning_rate": 1.4773960216998193e-05,
      "loss": 0.0013,
      "step": 3896
    },
    {
      "epoch": 4.93,
      "learning_rate": 1.4764918625678121e-05,
      "loss": 0.0127,
      "step": 3897
    },
    {
      "epoch": 4.93,
      "learning_rate": 1.4755877034358048e-05,
      "loss": 0.3944,
      "step": 3898
    },
    {
      "epoch": 4.93,
      "learning_rate": 1.4746835443037976e-05,
      "loss": 0.0687,
      "step": 3899
    },
    {
      "epoch": 4.94,
      "learning_rate": 1.4737793851717904e-05,
      "loss": 0.0013,
      "step": 3900
    },
    {
      "epoch": 4.94,
      "learning_rate": 1.4728752260397829e-05,
      "loss": 0.0004,
      "step": 3901
    },
    {
      "epoch": 4.94,
      "learning_rate": 1.4719710669077757e-05,
      "loss": 0.0342,
      "step": 3902
    },
    {
      "epoch": 4.94,
      "learning_rate": 1.4710669077757685e-05,
      "loss": 0.0009,
      "step": 3903
    },
    {
      "epoch": 4.94,
      "learning_rate": 1.4701627486437614e-05,
      "loss": 0.1422,
      "step": 3904
    },
    {
      "epoch": 4.94,
      "learning_rate": 1.4692585895117542e-05,
      "loss": 0.2238,
      "step": 3905
    },
    {
      "epoch": 4.94,
      "learning_rate": 1.468354430379747e-05,
      "loss": 0.0009,
      "step": 3906
    },
    {
      "epoch": 4.94,
      "learning_rate": 1.4674502712477398e-05,
      "loss": 0.0016,
      "step": 3907
    },
    {
      "epoch": 4.95,
      "learning_rate": 1.4665461121157325e-05,
      "loss": 0.0018,
      "step": 3908
    },
    {
      "epoch": 4.95,
      "learning_rate": 1.4656419529837251e-05,
      "loss": 0.0012,
      "step": 3909
    },
    {
      "epoch": 4.95,
      "learning_rate": 1.464737793851718e-05,
      "loss": 0.0566,
      "step": 3910
    },
    {
      "epoch": 4.95,
      "learning_rate": 1.4638336347197106e-05,
      "loss": 0.0024,
      "step": 3911
    },
    {
      "epoch": 4.95,
      "learning_rate": 1.4629294755877034e-05,
      "loss": 0.0005,
      "step": 3912
    },
    {
      "epoch": 4.95,
      "learning_rate": 1.4620253164556963e-05,
      "loss": 0.0011,
      "step": 3913
    },
    {
      "epoch": 4.95,
      "learning_rate": 1.4611211573236891e-05,
      "loss": 0.1444,
      "step": 3914
    },
    {
      "epoch": 4.96,
      "learning_rate": 1.4602169981916819e-05,
      "loss": 0.0058,
      "step": 3915
    },
    {
      "epoch": 4.96,
      "learning_rate": 1.4593128390596747e-05,
      "loss": 0.0022,
      "step": 3916
    },
    {
      "epoch": 4.96,
      "learning_rate": 1.4584086799276672e-05,
      "loss": 0.0272,
      "step": 3917
    },
    {
      "epoch": 4.96,
      "learning_rate": 1.45750452079566e-05,
      "loss": 0.0013,
      "step": 3918
    },
    {
      "epoch": 4.96,
      "learning_rate": 1.4566003616636529e-05,
      "loss": 0.0004,
      "step": 3919
    },
    {
      "epoch": 4.96,
      "learning_rate": 1.4556962025316457e-05,
      "loss": 0.0018,
      "step": 3920
    },
    {
      "epoch": 4.96,
      "learning_rate": 1.4547920433996383e-05,
      "loss": 0.0009,
      "step": 3921
    },
    {
      "epoch": 4.96,
      "learning_rate": 1.4538878842676312e-05,
      "loss": 0.0037,
      "step": 3922
    },
    {
      "epoch": 4.97,
      "learning_rate": 1.452983725135624e-05,
      "loss": 0.0323,
      "step": 3923
    },
    {
      "epoch": 4.97,
      "learning_rate": 1.4520795660036168e-05,
      "loss": 0.028,
      "step": 3924
    },
    {
      "epoch": 4.97,
      "learning_rate": 1.4511754068716093e-05,
      "loss": 0.0119,
      "step": 3925
    },
    {
      "epoch": 4.97,
      "learning_rate": 1.4502712477396021e-05,
      "loss": 0.0047,
      "step": 3926
    },
    {
      "epoch": 4.97,
      "learning_rate": 1.449367088607595e-05,
      "loss": 0.0013,
      "step": 3927
    },
    {
      "epoch": 4.97,
      "learning_rate": 1.4484629294755878e-05,
      "loss": 0.0008,
      "step": 3928
    },
    {
      "epoch": 4.97,
      "learning_rate": 1.4475587703435806e-05,
      "loss": 0.0064,
      "step": 3929
    },
    {
      "epoch": 4.97,
      "learning_rate": 1.4466546112115734e-05,
      "loss": 0.0007,
      "step": 3930
    },
    {
      "epoch": 4.98,
      "learning_rate": 1.4457504520795662e-05,
      "loss": 0.0057,
      "step": 3931
    },
    {
      "epoch": 4.98,
      "learning_rate": 1.4448462929475589e-05,
      "loss": 0.0008,
      "step": 3932
    },
    {
      "epoch": 4.98,
      "learning_rate": 1.4439421338155516e-05,
      "loss": 0.0089,
      "step": 3933
    },
    {
      "epoch": 4.98,
      "learning_rate": 1.4430379746835442e-05,
      "loss": 0.009,
      "step": 3934
    },
    {
      "epoch": 4.98,
      "learning_rate": 1.442133815551537e-05,
      "loss": 0.0175,
      "step": 3935
    },
    {
      "epoch": 4.98,
      "learning_rate": 1.4412296564195299e-05,
      "loss": 0.1225,
      "step": 3936
    },
    {
      "epoch": 4.98,
      "learning_rate": 1.4403254972875227e-05,
      "loss": 0.0009,
      "step": 3937
    },
    {
      "epoch": 4.98,
      "learning_rate": 1.4394213381555155e-05,
      "loss": 0.0009,
      "step": 3938
    },
    {
      "epoch": 4.99,
      "learning_rate": 1.4385171790235083e-05,
      "loss": 0.0027,
      "step": 3939
    },
    {
      "epoch": 4.99,
      "learning_rate": 1.4376130198915012e-05,
      "loss": 0.0947,
      "step": 3940
    },
    {
      "epoch": 4.99,
      "learning_rate": 1.4367088607594936e-05,
      "loss": 0.001,
      "step": 3941
    },
    {
      "epoch": 4.99,
      "learning_rate": 1.4358047016274865e-05,
      "loss": 0.0414,
      "step": 3942
    },
    {
      "epoch": 4.99,
      "learning_rate": 1.4349005424954793e-05,
      "loss": 0.0014,
      "step": 3943
    },
    {
      "epoch": 4.99,
      "learning_rate": 1.4339963833634721e-05,
      "loss": 0.051,
      "step": 3944
    },
    {
      "epoch": 4.99,
      "learning_rate": 1.4330922242314648e-05,
      "loss": 0.0069,
      "step": 3945
    },
    {
      "epoch": 4.99,
      "learning_rate": 1.4321880650994576e-05,
      "loss": 0.0486,
      "step": 3946
    },
    {
      "epoch": 5.0,
      "learning_rate": 1.4312839059674504e-05,
      "loss": 0.0492,
      "step": 3947
    },
    {
      "epoch": 5.0,
      "learning_rate": 1.4303797468354432e-05,
      "loss": 0.0244,
      "step": 3948
    },
    {
      "epoch": 5.0,
      "learning_rate": 1.4294755877034357e-05,
      "loss": 0.0124,
      "step": 3949
    },
    {
      "epoch": 5.0,
      "learning_rate": 1.4285714285714285e-05,
      "loss": 0.0012,
      "step": 3950
    },
    {
      "epoch": 5.0,
      "learning_rate": 1.4276672694394214e-05,
      "loss": 0.0013,
      "step": 3951
    },
    {
      "epoch": 5.0,
      "learning_rate": 1.4267631103074142e-05,
      "loss": 0.0015,
      "step": 3952
    },
    {
      "epoch": 5.0,
      "learning_rate": 1.425858951175407e-05,
      "loss": 0.0006,
      "step": 3953
    },
    {
      "epoch": 5.01,
      "learning_rate": 1.4249547920433998e-05,
      "loss": 0.0022,
      "step": 3954
    },
    {
      "epoch": 5.01,
      "learning_rate": 1.4240506329113925e-05,
      "loss": 0.0006,
      "step": 3955
    },
    {
      "epoch": 5.01,
      "learning_rate": 1.4231464737793853e-05,
      "loss": 0.0007,
      "step": 3956
    },
    {
      "epoch": 5.01,
      "learning_rate": 1.422242314647378e-05,
      "loss": 0.0017,
      "step": 3957
    },
    {
      "epoch": 5.01,
      "learning_rate": 1.4213381555153706e-05,
      "loss": 0.0008,
      "step": 3958
    },
    {
      "epoch": 5.01,
      "learning_rate": 1.4204339963833634e-05,
      "loss": 0.0005,
      "step": 3959
    },
    {
      "epoch": 5.01,
      "learning_rate": 1.4195298372513563e-05,
      "loss": 0.0027,
      "step": 3960
    },
    {
      "epoch": 5.01,
      "learning_rate": 1.4186256781193491e-05,
      "loss": 0.0006,
      "step": 3961
    },
    {
      "epoch": 5.02,
      "learning_rate": 1.417721518987342e-05,
      "loss": 0.0003,
      "step": 3962
    },
    {
      "epoch": 5.02,
      "learning_rate": 1.4168173598553347e-05,
      "loss": 0.0008,
      "step": 3963
    },
    {
      "epoch": 5.02,
      "learning_rate": 1.4159132007233276e-05,
      "loss": 0.0008,
      "step": 3964
    },
    {
      "epoch": 5.02,
      "learning_rate": 1.41500904159132e-05,
      "loss": 0.0006,
      "step": 3965
    },
    {
      "epoch": 5.02,
      "learning_rate": 1.4141048824593129e-05,
      "loss": 0.1093,
      "step": 3966
    },
    {
      "epoch": 5.02,
      "learning_rate": 1.4132007233273057e-05,
      "loss": 0.1127,
      "step": 3967
    },
    {
      "epoch": 5.02,
      "learning_rate": 1.4122965641952984e-05,
      "loss": 0.0008,
      "step": 3968
    },
    {
      "epoch": 5.02,
      "learning_rate": 1.4113924050632912e-05,
      "loss": 0.0044,
      "step": 3969
    },
    {
      "epoch": 5.03,
      "learning_rate": 1.410488245931284e-05,
      "loss": 0.0018,
      "step": 3970
    },
    {
      "epoch": 5.03,
      "learning_rate": 1.4095840867992768e-05,
      "loss": 0.0564,
      "step": 3971
    },
    {
      "epoch": 5.03,
      "learning_rate": 1.4086799276672696e-05,
      "loss": 0.0005,
      "step": 3972
    },
    {
      "epoch": 5.03,
      "learning_rate": 1.4077757685352621e-05,
      "loss": 0.0182,
      "step": 3973
    },
    {
      "epoch": 5.03,
      "learning_rate": 1.406871609403255e-05,
      "loss": 0.0013,
      "step": 3974
    },
    {
      "epoch": 5.03,
      "learning_rate": 1.4059674502712478e-05,
      "loss": 0.0182,
      "step": 3975
    },
    {
      "epoch": 5.03,
      "learning_rate": 1.4050632911392406e-05,
      "loss": 0.0004,
      "step": 3976
    },
    {
      "epoch": 5.03,
      "learning_rate": 1.4041591320072334e-05,
      "loss": 0.001,
      "step": 3977
    },
    {
      "epoch": 5.04,
      "learning_rate": 1.4032549728752263e-05,
      "loss": 0.0097,
      "step": 3978
    },
    {
      "epoch": 5.04,
      "learning_rate": 1.4023508137432189e-05,
      "loss": 0.0004,
      "step": 3979
    },
    {
      "epoch": 5.04,
      "learning_rate": 1.4014466546112117e-05,
      "loss": 0.0006,
      "step": 3980
    },
    {
      "epoch": 5.04,
      "learning_rate": 1.4005424954792042e-05,
      "loss": 0.0018,
      "step": 3981
    },
    {
      "epoch": 5.04,
      "learning_rate": 1.399638336347197e-05,
      "loss": 0.0007,
      "step": 3982
    },
    {
      "epoch": 5.04,
      "learning_rate": 1.3987341772151899e-05,
      "loss": 0.0007,
      "step": 3983
    },
    {
      "epoch": 5.04,
      "learning_rate": 1.3978300180831827e-05,
      "loss": 0.0008,
      "step": 3984
    },
    {
      "epoch": 5.04,
      "learning_rate": 1.3969258589511755e-05,
      "loss": 0.0123,
      "step": 3985
    },
    {
      "epoch": 5.05,
      "learning_rate": 1.3960216998191683e-05,
      "loss": 0.0018,
      "step": 3986
    },
    {
      "epoch": 5.05,
      "learning_rate": 1.3951175406871612e-05,
      "loss": 0.0024,
      "step": 3987
    },
    {
      "epoch": 5.05,
      "learning_rate": 1.394213381555154e-05,
      "loss": 0.001,
      "step": 3988
    },
    {
      "epoch": 5.05,
      "learning_rate": 1.3933092224231465e-05,
      "loss": 0.0004,
      "step": 3989
    },
    {
      "epoch": 5.05,
      "learning_rate": 1.3924050632911393e-05,
      "loss": 0.0011,
      "step": 3990
    },
    {
      "epoch": 5.05,
      "learning_rate": 1.3915009041591321e-05,
      "loss": 0.0183,
      "step": 3991
    },
    {
      "epoch": 5.05,
      "learning_rate": 1.3905967450271248e-05,
      "loss": 0.0004,
      "step": 3992
    },
    {
      "epoch": 5.05,
      "learning_rate": 1.3896925858951176e-05,
      "loss": 0.0029,
      "step": 3993
    },
    {
      "epoch": 5.06,
      "learning_rate": 1.3887884267631104e-05,
      "loss": 0.2541,
      "step": 3994
    },
    {
      "epoch": 5.06,
      "learning_rate": 1.3878842676311032e-05,
      "loss": 0.0009,
      "step": 3995
    },
    {
      "epoch": 5.06,
      "learning_rate": 1.386980108499096e-05,
      "loss": 0.0775,
      "step": 3996
    },
    {
      "epoch": 5.06,
      "learning_rate": 1.3860759493670886e-05,
      "loss": 0.0006,
      "step": 3997
    },
    {
      "epoch": 5.06,
      "learning_rate": 1.3851717902350814e-05,
      "loss": 0.001,
      "step": 3998
    },
    {
      "epoch": 5.06,
      "learning_rate": 1.3842676311030742e-05,
      "loss": 0.0006,
      "step": 3999
    },
    {
      "epoch": 5.06,
      "learning_rate": 1.383363471971067e-05,
      "loss": 0.0025,
      "step": 4000
    },
    {
      "epoch": 5.06,
      "eval_accuracy": 0.9054054054054054,
      "eval_loss": 0.4468367397785187,
      "eval_runtime": 45.1029,
      "eval_samples_per_second": 29.533,
      "eval_steps_per_second": 3.703,
      "step": 4000
    },
    {
      "epoch": 5.06,
      "learning_rate": 1.3824593128390598e-05,
      "loss": 0.0008,
      "step": 4001
    },
    {
      "epoch": 5.07,
      "learning_rate": 1.3815551537070525e-05,
      "loss": 0.0009,
      "step": 4002
    },
    {
      "epoch": 5.07,
      "learning_rate": 1.3806509945750453e-05,
      "loss": 0.0009,
      "step": 4003
    },
    {
      "epoch": 5.07,
      "learning_rate": 1.3797468354430381e-05,
      "loss": 0.0006,
      "step": 4004
    },
    {
      "epoch": 5.07,
      "learning_rate": 1.3788426763110306e-05,
      "loss": 0.0008,
      "step": 4005
    },
    {
      "epoch": 5.07,
      "learning_rate": 1.3779385171790235e-05,
      "loss": 0.001,
      "step": 4006
    },
    {
      "epoch": 5.07,
      "learning_rate": 1.3770343580470163e-05,
      "loss": 0.0005,
      "step": 4007
    },
    {
      "epoch": 5.07,
      "learning_rate": 1.3761301989150091e-05,
      "loss": 0.0032,
      "step": 4008
    },
    {
      "epoch": 5.07,
      "learning_rate": 1.375226039783002e-05,
      "loss": 0.0008,
      "step": 4009
    },
    {
      "epoch": 5.08,
      "learning_rate": 1.3743218806509948e-05,
      "loss": 0.0297,
      "step": 4010
    },
    {
      "epoch": 5.08,
      "learning_rate": 1.3734177215189876e-05,
      "loss": 0.0004,
      "step": 4011
    },
    {
      "epoch": 5.08,
      "learning_rate": 1.3725135623869802e-05,
      "loss": 0.1201,
      "step": 4012
    },
    {
      "epoch": 5.08,
      "learning_rate": 1.3716094032549729e-05,
      "loss": 0.0008,
      "step": 4013
    },
    {
      "epoch": 5.08,
      "learning_rate": 1.3707052441229657e-05,
      "loss": 0.0013,
      "step": 4014
    },
    {
      "epoch": 5.08,
      "learning_rate": 1.3698010849909584e-05,
      "loss": 0.001,
      "step": 4015
    },
    {
      "epoch": 5.08,
      "learning_rate": 1.3688969258589512e-05,
      "loss": 0.27,
      "step": 4016
    },
    {
      "epoch": 5.08,
      "learning_rate": 1.367992766726944e-05,
      "loss": 0.0029,
      "step": 4017
    },
    {
      "epoch": 5.09,
      "learning_rate": 1.3670886075949368e-05,
      "loss": 0.0009,
      "step": 4018
    },
    {
      "epoch": 5.09,
      "learning_rate": 1.3661844484629297e-05,
      "loss": 0.0023,
      "step": 4019
    },
    {
      "epoch": 5.09,
      "learning_rate": 1.3652802893309225e-05,
      "loss": 0.0019,
      "step": 4020
    },
    {
      "epoch": 5.09,
      "learning_rate": 1.364376130198915e-05,
      "loss": 0.1304,
      "step": 4021
    },
    {
      "epoch": 5.09,
      "learning_rate": 1.3634719710669078e-05,
      "loss": 0.0012,
      "step": 4022
    },
    {
      "epoch": 5.09,
      "learning_rate": 1.3625678119349006e-05,
      "loss": 0.0057,
      "step": 4023
    },
    {
      "epoch": 5.09,
      "learning_rate": 1.3616636528028934e-05,
      "loss": 0.0007,
      "step": 4024
    },
    {
      "epoch": 5.09,
      "learning_rate": 1.3607594936708861e-05,
      "loss": 0.001,
      "step": 4025
    },
    {
      "epoch": 5.1,
      "learning_rate": 1.359855334538879e-05,
      "loss": 0.0007,
      "step": 4026
    },
    {
      "epoch": 5.1,
      "learning_rate": 1.3589511754068717e-05,
      "loss": 0.1467,
      "step": 4027
    },
    {
      "epoch": 5.1,
      "learning_rate": 1.3580470162748642e-05,
      "loss": 0.0031,
      "step": 4028
    },
    {
      "epoch": 5.1,
      "learning_rate": 1.357142857142857e-05,
      "loss": 0.0917,
      "step": 4029
    },
    {
      "epoch": 5.1,
      "learning_rate": 1.3562386980108499e-05,
      "loss": 0.0006,
      "step": 4030
    },
    {
      "epoch": 5.1,
      "learning_rate": 1.3553345388788427e-05,
      "loss": 0.0041,
      "step": 4031
    },
    {
      "epoch": 5.1,
      "learning_rate": 1.3544303797468355e-05,
      "loss": 0.0024,
      "step": 4032
    },
    {
      "epoch": 5.1,
      "learning_rate": 1.3535262206148283e-05,
      "loss": 0.0006,
      "step": 4033
    },
    {
      "epoch": 5.11,
      "learning_rate": 1.3526220614828212e-05,
      "loss": 0.0069,
      "step": 4034
    },
    {
      "epoch": 5.11,
      "learning_rate": 1.351717902350814e-05,
      "loss": 0.0068,
      "step": 4035
    },
    {
      "epoch": 5.11,
      "learning_rate": 1.3508137432188065e-05,
      "loss": 0.0006,
      "step": 4036
    },
    {
      "epoch": 5.11,
      "learning_rate": 1.3499095840867993e-05,
      "loss": 0.001,
      "step": 4037
    },
    {
      "epoch": 5.11,
      "learning_rate": 1.3490054249547921e-05,
      "loss": 0.0009,
      "step": 4038
    },
    {
      "epoch": 5.11,
      "learning_rate": 1.3481012658227848e-05,
      "loss": 0.0136,
      "step": 4039
    },
    {
      "epoch": 5.11,
      "learning_rate": 1.3471971066907776e-05,
      "loss": 0.0027,
      "step": 4040
    },
    {
      "epoch": 5.12,
      "learning_rate": 1.3462929475587704e-05,
      "loss": 0.0016,
      "step": 4041
    },
    {
      "epoch": 5.12,
      "learning_rate": 1.3453887884267632e-05,
      "loss": 0.1104,
      "step": 4042
    },
    {
      "epoch": 5.12,
      "learning_rate": 1.344484629294756e-05,
      "loss": 0.0009,
      "step": 4043
    },
    {
      "epoch": 5.12,
      "learning_rate": 1.3435804701627486e-05,
      "loss": 0.0004,
      "step": 4044
    },
    {
      "epoch": 5.12,
      "learning_rate": 1.3426763110307414e-05,
      "loss": 0.0014,
      "step": 4045
    },
    {
      "epoch": 5.12,
      "learning_rate": 1.3417721518987342e-05,
      "loss": 0.0042,
      "step": 4046
    },
    {
      "epoch": 5.12,
      "learning_rate": 1.340867992766727e-05,
      "loss": 0.1609,
      "step": 4047
    },
    {
      "epoch": 5.12,
      "learning_rate": 1.3399638336347199e-05,
      "loss": 0.0006,
      "step": 4048
    },
    {
      "epoch": 5.13,
      "learning_rate": 1.3390596745027125e-05,
      "loss": 0.0006,
      "step": 4049
    },
    {
      "epoch": 5.13,
      "learning_rate": 1.3381555153707053e-05,
      "loss": 0.0007,
      "step": 4050
    },
    {
      "epoch": 5.13,
      "learning_rate": 1.3372513562386982e-05,
      "loss": 0.0004,
      "step": 4051
    },
    {
      "epoch": 5.13,
      "learning_rate": 1.3363471971066906e-05,
      "loss": 0.0009,
      "step": 4052
    },
    {
      "epoch": 5.13,
      "learning_rate": 1.3354430379746835e-05,
      "loss": 0.0008,
      "step": 4053
    },
    {
      "epoch": 5.13,
      "learning_rate": 1.3345388788426763e-05,
      "loss": 0.0007,
      "step": 4054
    },
    {
      "epoch": 5.13,
      "learning_rate": 1.3336347197106691e-05,
      "loss": 0.0008,
      "step": 4055
    },
    {
      "epoch": 5.13,
      "learning_rate": 1.332730560578662e-05,
      "loss": 0.0058,
      "step": 4056
    },
    {
      "epoch": 5.14,
      "learning_rate": 1.3318264014466548e-05,
      "loss": 0.0014,
      "step": 4057
    },
    {
      "epoch": 5.14,
      "learning_rate": 1.3309222423146476e-05,
      "loss": 0.0011,
      "step": 4058
    },
    {
      "epoch": 5.14,
      "learning_rate": 1.3300180831826402e-05,
      "loss": 0.0007,
      "step": 4059
    },
    {
      "epoch": 5.14,
      "learning_rate": 1.3291139240506329e-05,
      "loss": 0.0529,
      "step": 4060
    },
    {
      "epoch": 5.14,
      "learning_rate": 1.3282097649186257e-05,
      "loss": 0.0007,
      "step": 4061
    },
    {
      "epoch": 5.14,
      "learning_rate": 1.3273056057866184e-05,
      "loss": 0.0012,
      "step": 4062
    },
    {
      "epoch": 5.14,
      "learning_rate": 1.3264014466546112e-05,
      "loss": 0.2208,
      "step": 4063
    },
    {
      "epoch": 5.14,
      "learning_rate": 1.325497287522604e-05,
      "loss": 0.0157,
      "step": 4064
    },
    {
      "epoch": 5.15,
      "learning_rate": 1.3245931283905968e-05,
      "loss": 0.0064,
      "step": 4065
    },
    {
      "epoch": 5.15,
      "learning_rate": 1.3236889692585897e-05,
      "loss": 0.0013,
      "step": 4066
    },
    {
      "epoch": 5.15,
      "learning_rate": 1.3227848101265825e-05,
      "loss": 0.2489,
      "step": 4067
    },
    {
      "epoch": 5.15,
      "learning_rate": 1.321880650994575e-05,
      "loss": 0.0006,
      "step": 4068
    },
    {
      "epoch": 5.15,
      "learning_rate": 1.3209764918625678e-05,
      "loss": 0.0018,
      "step": 4069
    },
    {
      "epoch": 5.15,
      "learning_rate": 1.3200723327305606e-05,
      "loss": 0.0005,
      "step": 4070
    },
    {
      "epoch": 5.15,
      "learning_rate": 1.3191681735985534e-05,
      "loss": 0.0005,
      "step": 4071
    },
    {
      "epoch": 5.15,
      "learning_rate": 1.3182640144665461e-05,
      "loss": 0.0008,
      "step": 4072
    },
    {
      "epoch": 5.16,
      "learning_rate": 1.317359855334539e-05,
      "loss": 0.0007,
      "step": 4073
    },
    {
      "epoch": 5.16,
      "learning_rate": 1.3164556962025317e-05,
      "loss": 0.001,
      "step": 4074
    },
    {
      "epoch": 5.16,
      "learning_rate": 1.3155515370705246e-05,
      "loss": 0.0006,
      "step": 4075
    },
    {
      "epoch": 5.16,
      "learning_rate": 1.314647377938517e-05,
      "loss": 0.0023,
      "step": 4076
    },
    {
      "epoch": 5.16,
      "learning_rate": 1.3137432188065099e-05,
      "loss": 0.0108,
      "step": 4077
    },
    {
      "epoch": 5.16,
      "learning_rate": 1.3128390596745027e-05,
      "loss": 0.0006,
      "step": 4078
    },
    {
      "epoch": 5.16,
      "learning_rate": 1.3119349005424955e-05,
      "loss": 0.0009,
      "step": 4079
    },
    {
      "epoch": 5.16,
      "learning_rate": 1.3110307414104884e-05,
      "loss": 0.0008,
      "step": 4080
    },
    {
      "epoch": 5.17,
      "learning_rate": 1.3101265822784812e-05,
      "loss": 0.0007,
      "step": 4081
    },
    {
      "epoch": 5.17,
      "learning_rate": 1.309222423146474e-05,
      "loss": 0.0866,
      "step": 4082
    },
    {
      "epoch": 5.17,
      "learning_rate": 1.3083182640144667e-05,
      "loss": 0.0037,
      "step": 4083
    },
    {
      "epoch": 5.17,
      "learning_rate": 1.3074141048824593e-05,
      "loss": 0.0004,
      "step": 4084
    },
    {
      "epoch": 5.17,
      "learning_rate": 1.3065099457504521e-05,
      "loss": 0.0015,
      "step": 4085
    },
    {
      "epoch": 5.17,
      "learning_rate": 1.3056057866184448e-05,
      "loss": 0.0247,
      "step": 4086
    },
    {
      "epoch": 5.17,
      "learning_rate": 1.3047016274864376e-05,
      "loss": 0.0015,
      "step": 4087
    },
    {
      "epoch": 5.17,
      "learning_rate": 1.3037974683544304e-05,
      "loss": 0.0006,
      "step": 4088
    },
    {
      "epoch": 5.18,
      "learning_rate": 1.3028933092224233e-05,
      "loss": 0.0004,
      "step": 4089
    },
    {
      "epoch": 5.18,
      "learning_rate": 1.301989150090416e-05,
      "loss": 0.0007,
      "step": 4090
    },
    {
      "epoch": 5.18,
      "learning_rate": 1.3010849909584089e-05,
      "loss": 0.001,
      "step": 4091
    },
    {
      "epoch": 5.18,
      "learning_rate": 1.3001808318264014e-05,
      "loss": 0.0009,
      "step": 4092
    },
    {
      "epoch": 5.18,
      "learning_rate": 1.2992766726943942e-05,
      "loss": 0.001,
      "step": 4093
    },
    {
      "epoch": 5.18,
      "learning_rate": 1.298372513562387e-05,
      "loss": 0.0024,
      "step": 4094
    },
    {
      "epoch": 5.18,
      "learning_rate": 1.2974683544303799e-05,
      "loss": 0.0009,
      "step": 4095
    },
    {
      "epoch": 5.18,
      "learning_rate": 1.2965641952983725e-05,
      "loss": 0.0007,
      "step": 4096
    },
    {
      "epoch": 5.19,
      "learning_rate": 1.2956600361663653e-05,
      "loss": 0.0008,
      "step": 4097
    },
    {
      "epoch": 5.19,
      "learning_rate": 1.2947558770343582e-05,
      "loss": 0.0008,
      "step": 4098
    },
    {
      "epoch": 5.19,
      "learning_rate": 1.293851717902351e-05,
      "loss": 0.0009,
      "step": 4099
    },
    {
      "epoch": 5.19,
      "learning_rate": 1.2929475587703435e-05,
      "loss": 0.0027,
      "step": 4100
    },
    {
      "epoch": 5.19,
      "learning_rate": 1.2920433996383363e-05,
      "loss": 0.0009,
      "step": 4101
    },
    {
      "epoch": 5.19,
      "learning_rate": 1.2911392405063291e-05,
      "loss": 0.0326,
      "step": 4102
    },
    {
      "epoch": 5.19,
      "learning_rate": 1.290235081374322e-05,
      "loss": 0.0984,
      "step": 4103
    },
    {
      "epoch": 5.19,
      "learning_rate": 1.2893309222423148e-05,
      "loss": 0.0005,
      "step": 4104
    },
    {
      "epoch": 5.2,
      "learning_rate": 1.2884267631103076e-05,
      "loss": 0.0009,
      "step": 4105
    },
    {
      "epoch": 5.2,
      "learning_rate": 1.2875226039783002e-05,
      "loss": 0.0019,
      "step": 4106
    },
    {
      "epoch": 5.2,
      "learning_rate": 1.286618444846293e-05,
      "loss": 0.0004,
      "step": 4107
    },
    {
      "epoch": 5.2,
      "learning_rate": 1.2857142857142857e-05,
      "loss": 0.0011,
      "step": 4108
    },
    {
      "epoch": 5.2,
      "learning_rate": 1.2848101265822784e-05,
      "loss": 0.0019,
      "step": 4109
    },
    {
      "epoch": 5.2,
      "learning_rate": 1.2839059674502712e-05,
      "loss": 0.023,
      "step": 4110
    },
    {
      "epoch": 5.2,
      "learning_rate": 1.283001808318264e-05,
      "loss": 0.2219,
      "step": 4111
    },
    {
      "epoch": 5.2,
      "learning_rate": 1.2820976491862568e-05,
      "loss": 0.0003,
      "step": 4112
    },
    {
      "epoch": 5.21,
      "learning_rate": 1.2811934900542497e-05,
      "loss": 0.0008,
      "step": 4113
    },
    {
      "epoch": 5.21,
      "learning_rate": 1.2802893309222425e-05,
      "loss": 0.0004,
      "step": 4114
    },
    {
      "epoch": 5.21,
      "learning_rate": 1.2793851717902353e-05,
      "loss": 0.0032,
      "step": 4115
    },
    {
      "epoch": 5.21,
      "learning_rate": 1.2784810126582278e-05,
      "loss": 0.0039,
      "step": 4116
    },
    {
      "epoch": 5.21,
      "learning_rate": 1.2775768535262206e-05,
      "loss": 0.001,
      "step": 4117
    },
    {
      "epoch": 5.21,
      "learning_rate": 1.2766726943942135e-05,
      "loss": 0.0005,
      "step": 4118
    },
    {
      "epoch": 5.21,
      "learning_rate": 1.2757685352622061e-05,
      "loss": 0.001,
      "step": 4119
    },
    {
      "epoch": 5.22,
      "learning_rate": 1.274864376130199e-05,
      "loss": 0.0008,
      "step": 4120
    },
    {
      "epoch": 5.22,
      "learning_rate": 1.2739602169981918e-05,
      "loss": 0.0013,
      "step": 4121
    },
    {
      "epoch": 5.22,
      "learning_rate": 1.2730560578661846e-05,
      "loss": 0.0008,
      "step": 4122
    },
    {
      "epoch": 5.22,
      "learning_rate": 1.2721518987341774e-05,
      "loss": 0.001,
      "step": 4123
    },
    {
      "epoch": 5.22,
      "learning_rate": 1.2712477396021699e-05,
      "loss": 0.0015,
      "step": 4124
    },
    {
      "epoch": 5.22,
      "learning_rate": 1.2703435804701627e-05,
      "loss": 0.2518,
      "step": 4125
    },
    {
      "epoch": 5.22,
      "learning_rate": 1.2694394213381555e-05,
      "loss": 0.0043,
      "step": 4126
    },
    {
      "epoch": 5.22,
      "learning_rate": 1.2685352622061484e-05,
      "loss": 0.0014,
      "step": 4127
    },
    {
      "epoch": 5.23,
      "learning_rate": 1.2676311030741412e-05,
      "loss": 0.0006,
      "step": 4128
    },
    {
      "epoch": 5.23,
      "learning_rate": 1.266726943942134e-05,
      "loss": 0.0021,
      "step": 4129
    },
    {
      "epoch": 5.23,
      "learning_rate": 1.2658227848101267e-05,
      "loss": 0.0007,
      "step": 4130
    },
    {
      "epoch": 5.23,
      "learning_rate": 1.2649186256781195e-05,
      "loss": 0.0011,
      "step": 4131
    },
    {
      "epoch": 5.23,
      "learning_rate": 1.264014466546112e-05,
      "loss": 0.0007,
      "step": 4132
    },
    {
      "epoch": 5.23,
      "learning_rate": 1.2631103074141048e-05,
      "loss": 0.0006,
      "step": 4133
    },
    {
      "epoch": 5.23,
      "learning_rate": 1.2622061482820976e-05,
      "loss": 0.0006,
      "step": 4134
    },
    {
      "epoch": 5.23,
      "learning_rate": 1.2613019891500904e-05,
      "loss": 0.0007,
      "step": 4135
    },
    {
      "epoch": 5.24,
      "learning_rate": 1.2603978300180833e-05,
      "loss": 0.0006,
      "step": 4136
    },
    {
      "epoch": 5.24,
      "learning_rate": 1.2594936708860761e-05,
      "loss": 0.0049,
      "step": 4137
    },
    {
      "epoch": 5.24,
      "learning_rate": 1.2585895117540689e-05,
      "loss": 0.0008,
      "step": 4138
    },
    {
      "epoch": 5.24,
      "learning_rate": 1.2576853526220617e-05,
      "loss": 0.0007,
      "step": 4139
    },
    {
      "epoch": 5.24,
      "learning_rate": 1.2567811934900542e-05,
      "loss": 0.0017,
      "step": 4140
    },
    {
      "epoch": 5.24,
      "learning_rate": 1.255877034358047e-05,
      "loss": 0.0052,
      "step": 4141
    },
    {
      "epoch": 5.24,
      "learning_rate": 1.2549728752260399e-05,
      "loss": 0.001,
      "step": 4142
    },
    {
      "epoch": 5.24,
      "learning_rate": 1.2540687160940325e-05,
      "loss": 0.0625,
      "step": 4143
    },
    {
      "epoch": 5.25,
      "learning_rate": 1.2531645569620253e-05,
      "loss": 0.0007,
      "step": 4144
    },
    {
      "epoch": 5.25,
      "learning_rate": 1.2522603978300182e-05,
      "loss": 0.0068,
      "step": 4145
    },
    {
      "epoch": 5.25,
      "learning_rate": 1.251356238698011e-05,
      "loss": 0.0017,
      "step": 4146
    },
    {
      "epoch": 5.25,
      "learning_rate": 1.2504520795660038e-05,
      "loss": 0.0004,
      "step": 4147
    },
    {
      "epoch": 5.25,
      "learning_rate": 1.2495479204339965e-05,
      "loss": 0.009,
      "step": 4148
    },
    {
      "epoch": 5.25,
      "learning_rate": 1.2486437613019893e-05,
      "loss": 0.0014,
      "step": 4149
    },
    {
      "epoch": 5.25,
      "learning_rate": 1.247739602169982e-05,
      "loss": 0.0014,
      "step": 4150
    },
    {
      "epoch": 5.25,
      "learning_rate": 1.2468354430379748e-05,
      "loss": 0.0006,
      "step": 4151
    },
    {
      "epoch": 5.26,
      "learning_rate": 1.2459312839059676e-05,
      "loss": 0.0025,
      "step": 4152
    },
    {
      "epoch": 5.26,
      "learning_rate": 1.2450271247739603e-05,
      "loss": 0.001,
      "step": 4153
    },
    {
      "epoch": 5.26,
      "learning_rate": 1.244122965641953e-05,
      "loss": 0.0011,
      "step": 4154
    },
    {
      "epoch": 5.26,
      "learning_rate": 1.2432188065099457e-05,
      "loss": 0.041,
      "step": 4155
    },
    {
      "epoch": 5.26,
      "learning_rate": 1.2423146473779386e-05,
      "loss": 0.0006,
      "step": 4156
    },
    {
      "epoch": 5.26,
      "learning_rate": 1.2414104882459314e-05,
      "loss": 0.229,
      "step": 4157
    },
    {
      "epoch": 5.26,
      "learning_rate": 1.240506329113924e-05,
      "loss": 0.1389,
      "step": 4158
    },
    {
      "epoch": 5.26,
      "learning_rate": 1.2396021699819169e-05,
      "loss": 0.0012,
      "step": 4159
    },
    {
      "epoch": 5.27,
      "learning_rate": 1.2386980108499097e-05,
      "loss": 0.0331,
      "step": 4160
    },
    {
      "epoch": 5.27,
      "learning_rate": 1.2377938517179025e-05,
      "loss": 0.0739,
      "step": 4161
    },
    {
      "epoch": 5.27,
      "learning_rate": 1.2368896925858952e-05,
      "loss": 0.001,
      "step": 4162
    },
    {
      "epoch": 5.27,
      "learning_rate": 1.235985533453888e-05,
      "loss": 0.0004,
      "step": 4163
    },
    {
      "epoch": 5.27,
      "learning_rate": 1.2350813743218808e-05,
      "loss": 0.044,
      "step": 4164
    },
    {
      "epoch": 5.27,
      "learning_rate": 1.2341772151898735e-05,
      "loss": 0.0023,
      "step": 4165
    },
    {
      "epoch": 5.27,
      "learning_rate": 1.2332730560578661e-05,
      "loss": 0.0407,
      "step": 4166
    },
    {
      "epoch": 5.27,
      "learning_rate": 1.232368896925859e-05,
      "loss": 0.0026,
      "step": 4167
    },
    {
      "epoch": 5.28,
      "learning_rate": 1.2314647377938518e-05,
      "loss": 0.0006,
      "step": 4168
    },
    {
      "epoch": 5.28,
      "learning_rate": 1.2305605786618446e-05,
      "loss": 0.0009,
      "step": 4169
    },
    {
      "epoch": 5.28,
      "learning_rate": 1.2296564195298372e-05,
      "loss": 0.0033,
      "step": 4170
    },
    {
      "epoch": 5.28,
      "learning_rate": 1.22875226039783e-05,
      "loss": 0.0007,
      "step": 4171
    },
    {
      "epoch": 5.28,
      "learning_rate": 1.2278481012658229e-05,
      "loss": 0.061,
      "step": 4172
    },
    {
      "epoch": 5.28,
      "learning_rate": 1.2269439421338157e-05,
      "loss": 0.0016,
      "step": 4173
    },
    {
      "epoch": 5.28,
      "learning_rate": 1.2260397830018084e-05,
      "loss": 0.0007,
      "step": 4174
    },
    {
      "epoch": 5.28,
      "learning_rate": 1.2251356238698012e-05,
      "loss": 0.0212,
      "step": 4175
    },
    {
      "epoch": 5.29,
      "learning_rate": 1.224231464737794e-05,
      "loss": 0.0026,
      "step": 4176
    },
    {
      "epoch": 5.29,
      "learning_rate": 1.2233273056057867e-05,
      "loss": 0.0015,
      "step": 4177
    },
    {
      "epoch": 5.29,
      "learning_rate": 1.2224231464737793e-05,
      "loss": 0.0008,
      "step": 4178
    },
    {
      "epoch": 5.29,
      "learning_rate": 1.2215189873417721e-05,
      "loss": 0.0012,
      "step": 4179
    },
    {
      "epoch": 5.29,
      "learning_rate": 1.220614828209765e-05,
      "loss": 0.0228,
      "step": 4180
    },
    {
      "epoch": 5.29,
      "learning_rate": 1.2197106690777578e-05,
      "loss": 0.0104,
      "step": 4181
    },
    {
      "epoch": 5.29,
      "learning_rate": 1.2188065099457504e-05,
      "loss": 0.0014,
      "step": 4182
    },
    {
      "epoch": 5.29,
      "learning_rate": 1.2179023508137433e-05,
      "loss": 0.0008,
      "step": 4183
    },
    {
      "epoch": 5.3,
      "learning_rate": 1.2169981916817361e-05,
      "loss": 0.0474,
      "step": 4184
    },
    {
      "epoch": 5.3,
      "learning_rate": 1.216094032549729e-05,
      "loss": 0.0012,
      "step": 4185
    },
    {
      "epoch": 5.3,
      "learning_rate": 1.2151898734177216e-05,
      "loss": 0.1058,
      "step": 4186
    },
    {
      "epoch": 5.3,
      "learning_rate": 1.2142857142857144e-05,
      "loss": 0.0008,
      "step": 4187
    },
    {
      "epoch": 5.3,
      "learning_rate": 1.213381555153707e-05,
      "loss": 0.0011,
      "step": 4188
    },
    {
      "epoch": 5.3,
      "learning_rate": 1.2124773960216999e-05,
      "loss": 0.0006,
      "step": 4189
    },
    {
      "epoch": 5.3,
      "learning_rate": 1.2115732368896925e-05,
      "loss": 0.0008,
      "step": 4190
    },
    {
      "epoch": 5.3,
      "learning_rate": 1.2106690777576854e-05,
      "loss": 0.0004,
      "step": 4191
    },
    {
      "epoch": 5.31,
      "learning_rate": 1.2097649186256782e-05,
      "loss": 0.0008,
      "step": 4192
    },
    {
      "epoch": 5.31,
      "learning_rate": 1.208860759493671e-05,
      "loss": 0.0011,
      "step": 4193
    },
    {
      "epoch": 5.31,
      "learning_rate": 1.2079566003616637e-05,
      "loss": 0.0037,
      "step": 4194
    },
    {
      "epoch": 5.31,
      "learning_rate": 1.2070524412296565e-05,
      "loss": 0.0681,
      "step": 4195
    },
    {
      "epoch": 5.31,
      "learning_rate": 1.2061482820976493e-05,
      "loss": 0.0574,
      "step": 4196
    },
    {
      "epoch": 5.31,
      "learning_rate": 1.2052441229656421e-05,
      "loss": 0.0043,
      "step": 4197
    },
    {
      "epoch": 5.31,
      "learning_rate": 1.2043399638336348e-05,
      "loss": 0.0009,
      "step": 4198
    },
    {
      "epoch": 5.31,
      "learning_rate": 1.2034358047016276e-05,
      "loss": 0.0078,
      "step": 4199
    },
    {
      "epoch": 5.32,
      "learning_rate": 1.2025316455696203e-05,
      "loss": 0.0005,
      "step": 4200
    },
    {
      "epoch": 5.32,
      "learning_rate": 1.201627486437613e-05,
      "loss": 0.082,
      "step": 4201
    },
    {
      "epoch": 5.32,
      "learning_rate": 1.2007233273056057e-05,
      "loss": 0.0012,
      "step": 4202
    },
    {
      "epoch": 5.32,
      "learning_rate": 1.1998191681735986e-05,
      "loss": 0.0331,
      "step": 4203
    },
    {
      "epoch": 5.32,
      "learning_rate": 1.1989150090415914e-05,
      "loss": 0.0012,
      "step": 4204
    },
    {
      "epoch": 5.32,
      "learning_rate": 1.1980108499095842e-05,
      "loss": 0.0012,
      "step": 4205
    },
    {
      "epoch": 5.32,
      "learning_rate": 1.1971066907775769e-05,
      "loss": 0.0007,
      "step": 4206
    },
    {
      "epoch": 5.33,
      "learning_rate": 1.1962025316455697e-05,
      "loss": 0.0029,
      "step": 4207
    },
    {
      "epoch": 5.33,
      "learning_rate": 1.1952983725135625e-05,
      "loss": 0.0028,
      "step": 4208
    },
    {
      "epoch": 5.33,
      "learning_rate": 1.1943942133815553e-05,
      "loss": 0.0022,
      "step": 4209
    },
    {
      "epoch": 5.33,
      "learning_rate": 1.193490054249548e-05,
      "loss": 0.0007,
      "step": 4210
    },
    {
      "epoch": 5.33,
      "learning_rate": 1.1925858951175408e-05,
      "loss": 0.0006,
      "step": 4211
    },
    {
      "epoch": 5.33,
      "learning_rate": 1.1916817359855335e-05,
      "loss": 0.0489,
      "step": 4212
    },
    {
      "epoch": 5.33,
      "learning_rate": 1.1907775768535263e-05,
      "loss": 0.0037,
      "step": 4213
    },
    {
      "epoch": 5.33,
      "learning_rate": 1.189873417721519e-05,
      "loss": 0.0759,
      "step": 4214
    },
    {
      "epoch": 5.34,
      "learning_rate": 1.1889692585895118e-05,
      "loss": 0.0033,
      "step": 4215
    },
    {
      "epoch": 5.34,
      "learning_rate": 1.1880650994575046e-05,
      "loss": 0.0004,
      "step": 4216
    },
    {
      "epoch": 5.34,
      "learning_rate": 1.1871609403254974e-05,
      "loss": 0.0041,
      "step": 4217
    },
    {
      "epoch": 5.34,
      "learning_rate": 1.18625678119349e-05,
      "loss": 0.0027,
      "step": 4218
    },
    {
      "epoch": 5.34,
      "learning_rate": 1.1853526220614829e-05,
      "loss": 0.066,
      "step": 4219
    },
    {
      "epoch": 5.34,
      "learning_rate": 1.1844484629294757e-05,
      "loss": 0.0011,
      "step": 4220
    },
    {
      "epoch": 5.34,
      "learning_rate": 1.1835443037974685e-05,
      "loss": 0.0006,
      "step": 4221
    },
    {
      "epoch": 5.34,
      "learning_rate": 1.1826401446654612e-05,
      "loss": 0.0009,
      "step": 4222
    },
    {
      "epoch": 5.35,
      "learning_rate": 1.181735985533454e-05,
      "loss": 0.0005,
      "step": 4223
    },
    {
      "epoch": 5.35,
      "learning_rate": 1.1808318264014467e-05,
      "loss": 0.0012,
      "step": 4224
    },
    {
      "epoch": 5.35,
      "learning_rate": 1.1799276672694395e-05,
      "loss": 0.0004,
      "step": 4225
    },
    {
      "epoch": 5.35,
      "learning_rate": 1.1790235081374322e-05,
      "loss": 0.0194,
      "step": 4226
    },
    {
      "epoch": 5.35,
      "learning_rate": 1.178119349005425e-05,
      "loss": 0.0015,
      "step": 4227
    },
    {
      "epoch": 5.35,
      "learning_rate": 1.1772151898734178e-05,
      "loss": 0.0006,
      "step": 4228
    },
    {
      "epoch": 5.35,
      "learning_rate": 1.1763110307414106e-05,
      "loss": 0.0005,
      "step": 4229
    },
    {
      "epoch": 5.35,
      "learning_rate": 1.1754068716094033e-05,
      "loss": 0.0061,
      "step": 4230
    },
    {
      "epoch": 5.36,
      "learning_rate": 1.1745027124773961e-05,
      "loss": 0.0021,
      "step": 4231
    },
    {
      "epoch": 5.36,
      "learning_rate": 1.173598553345389e-05,
      "loss": 0.003,
      "step": 4232
    },
    {
      "epoch": 5.36,
      "learning_rate": 1.1726943942133818e-05,
      "loss": 0.0019,
      "step": 4233
    },
    {
      "epoch": 5.36,
      "learning_rate": 1.1717902350813744e-05,
      "loss": 0.0008,
      "step": 4234
    },
    {
      "epoch": 5.36,
      "learning_rate": 1.170886075949367e-05,
      "loss": 0.0593,
      "step": 4235
    },
    {
      "epoch": 5.36,
      "learning_rate": 1.1699819168173599e-05,
      "loss": 0.0006,
      "step": 4236
    },
    {
      "epoch": 5.36,
      "learning_rate": 1.1690777576853527e-05,
      "loss": 0.0008,
      "step": 4237
    },
    {
      "epoch": 5.36,
      "learning_rate": 1.1681735985533454e-05,
      "loss": 0.0017,
      "step": 4238
    },
    {
      "epoch": 5.37,
      "learning_rate": 1.1672694394213382e-05,
      "loss": 0.0021,
      "step": 4239
    },
    {
      "epoch": 5.37,
      "learning_rate": 1.166365280289331e-05,
      "loss": 0.0005,
      "step": 4240
    },
    {
      "epoch": 5.37,
      "learning_rate": 1.1654611211573238e-05,
      "loss": 0.0008,
      "step": 4241
    },
    {
      "epoch": 5.37,
      "learning_rate": 1.1645569620253165e-05,
      "loss": 0.0012,
      "step": 4242
    },
    {
      "epoch": 5.37,
      "learning_rate": 1.1636528028933093e-05,
      "loss": 0.0015,
      "step": 4243
    },
    {
      "epoch": 5.37,
      "learning_rate": 1.1627486437613021e-05,
      "loss": 0.0005,
      "step": 4244
    },
    {
      "epoch": 5.37,
      "learning_rate": 1.161844484629295e-05,
      "loss": 0.0057,
      "step": 4245
    },
    {
      "epoch": 5.37,
      "learning_rate": 1.1609403254972876e-05,
      "loss": 0.0013,
      "step": 4246
    },
    {
      "epoch": 5.38,
      "learning_rate": 1.1600361663652803e-05,
      "loss": 0.0184,
      "step": 4247
    },
    {
      "epoch": 5.38,
      "learning_rate": 1.1591320072332731e-05,
      "loss": 0.0007,
      "step": 4248
    },
    {
      "epoch": 5.38,
      "learning_rate": 1.1582278481012659e-05,
      "loss": 0.0005,
      "step": 4249
    },
    {
      "epoch": 5.38,
      "learning_rate": 1.1573236889692586e-05,
      "loss": 0.0024,
      "step": 4250
    },
    {
      "epoch": 5.38,
      "eval_accuracy": 0.9076576576576577,
      "eval_loss": 0.4512460231781006,
      "eval_runtime": 45.103,
      "eval_samples_per_second": 29.532,
      "eval_steps_per_second": 3.703,
      "step": 4250
    },
    {
      "epoch": 5.38,
      "learning_rate": 1.1564195298372514e-05,
      "loss": 0.0016,
      "step": 4251
    },
    {
      "epoch": 5.38,
      "learning_rate": 1.1555153707052442e-05,
      "loss": 0.0008,
      "step": 4252
    },
    {
      "epoch": 5.38,
      "learning_rate": 1.154611211573237e-05,
      "loss": 0.0019,
      "step": 4253
    },
    {
      "epoch": 5.38,
      "learning_rate": 1.1537070524412297e-05,
      "loss": 0.0005,
      "step": 4254
    },
    {
      "epoch": 5.39,
      "learning_rate": 1.1528028933092225e-05,
      "loss": 0.0009,
      "step": 4255
    },
    {
      "epoch": 5.39,
      "learning_rate": 1.1518987341772153e-05,
      "loss": 0.1738,
      "step": 4256
    },
    {
      "epoch": 5.39,
      "learning_rate": 1.150994575045208e-05,
      "loss": 0.0006,
      "step": 4257
    },
    {
      "epoch": 5.39,
      "learning_rate": 1.1500904159132008e-05,
      "loss": 0.0004,
      "step": 4258
    },
    {
      "epoch": 5.39,
      "learning_rate": 1.1491862567811935e-05,
      "loss": 0.0046,
      "step": 4259
    },
    {
      "epoch": 5.39,
      "learning_rate": 1.1482820976491863e-05,
      "loss": 0.0005,
      "step": 4260
    },
    {
      "epoch": 5.39,
      "learning_rate": 1.1473779385171791e-05,
      "loss": 0.0031,
      "step": 4261
    },
    {
      "epoch": 5.39,
      "learning_rate": 1.1464737793851718e-05,
      "loss": 0.2014,
      "step": 4262
    },
    {
      "epoch": 5.4,
      "learning_rate": 1.1455696202531646e-05,
      "loss": 0.0009,
      "step": 4263
    },
    {
      "epoch": 5.4,
      "learning_rate": 1.1446654611211574e-05,
      "loss": 0.0056,
      "step": 4264
    },
    {
      "epoch": 5.4,
      "learning_rate": 1.1437613019891502e-05,
      "loss": 0.0007,
      "step": 4265
    },
    {
      "epoch": 5.4,
      "learning_rate": 1.1428571428571429e-05,
      "loss": 0.0008,
      "step": 4266
    },
    {
      "epoch": 5.4,
      "learning_rate": 1.1419529837251357e-05,
      "loss": 0.0091,
      "step": 4267
    },
    {
      "epoch": 5.4,
      "learning_rate": 1.1410488245931285e-05,
      "loss": 0.0006,
      "step": 4268
    },
    {
      "epoch": 5.4,
      "learning_rate": 1.1401446654611212e-05,
      "loss": 0.0013,
      "step": 4269
    },
    {
      "epoch": 5.4,
      "learning_rate": 1.139240506329114e-05,
      "loss": 0.0015,
      "step": 4270
    },
    {
      "epoch": 5.41,
      "learning_rate": 1.1383363471971067e-05,
      "loss": 0.0005,
      "step": 4271
    },
    {
      "epoch": 5.41,
      "learning_rate": 1.1374321880650995e-05,
      "loss": 0.0008,
      "step": 4272
    },
    {
      "epoch": 5.41,
      "learning_rate": 1.1365280289330923e-05,
      "loss": 0.076,
      "step": 4273
    },
    {
      "epoch": 5.41,
      "learning_rate": 1.135623869801085e-05,
      "loss": 0.0009,
      "step": 4274
    },
    {
      "epoch": 5.41,
      "learning_rate": 1.1347197106690778e-05,
      "loss": 0.0061,
      "step": 4275
    },
    {
      "epoch": 5.41,
      "learning_rate": 1.1338155515370706e-05,
      "loss": 0.0006,
      "step": 4276
    },
    {
      "epoch": 5.41,
      "learning_rate": 1.1329113924050633e-05,
      "loss": 0.0027,
      "step": 4277
    },
    {
      "epoch": 5.41,
      "learning_rate": 1.1320072332730561e-05,
      "loss": 0.005,
      "step": 4278
    },
    {
      "epoch": 5.42,
      "learning_rate": 1.131103074141049e-05,
      "loss": 0.0024,
      "step": 4279
    },
    {
      "epoch": 5.42,
      "learning_rate": 1.1301989150090418e-05,
      "loss": 0.0006,
      "step": 4280
    },
    {
      "epoch": 5.42,
      "learning_rate": 1.1292947558770344e-05,
      "loss": 0.013,
      "step": 4281
    },
    {
      "epoch": 5.42,
      "learning_rate": 1.128390596745027e-05,
      "loss": 0.0009,
      "step": 4282
    },
    {
      "epoch": 5.42,
      "learning_rate": 1.1274864376130199e-05,
      "loss": 0.0004,
      "step": 4283
    },
    {
      "epoch": 5.42,
      "learning_rate": 1.1265822784810127e-05,
      "loss": 0.0655,
      "step": 4284
    },
    {
      "epoch": 5.42,
      "learning_rate": 1.1256781193490054e-05,
      "loss": 0.0018,
      "step": 4285
    },
    {
      "epoch": 5.43,
      "learning_rate": 1.1247739602169982e-05,
      "loss": 0.0031,
      "step": 4286
    },
    {
      "epoch": 5.43,
      "learning_rate": 1.123869801084991e-05,
      "loss": 0.1632,
      "step": 4287
    },
    {
      "epoch": 5.43,
      "learning_rate": 1.1229656419529838e-05,
      "loss": 0.0012,
      "step": 4288
    },
    {
      "epoch": 5.43,
      "learning_rate": 1.1220614828209765e-05,
      "loss": 0.0032,
      "step": 4289
    },
    {
      "epoch": 5.43,
      "learning_rate": 1.1211573236889693e-05,
      "loss": 0.0008,
      "step": 4290
    },
    {
      "epoch": 5.43,
      "learning_rate": 1.1202531645569621e-05,
      "loss": 0.0005,
      "step": 4291
    },
    {
      "epoch": 5.43,
      "learning_rate": 1.119349005424955e-05,
      "loss": 0.1558,
      "step": 4292
    },
    {
      "epoch": 5.43,
      "learning_rate": 1.1184448462929476e-05,
      "loss": 0.0011,
      "step": 4293
    },
    {
      "epoch": 5.44,
      "learning_rate": 1.1175406871609403e-05,
      "loss": 0.0008,
      "step": 4294
    },
    {
      "epoch": 5.44,
      "learning_rate": 1.1166365280289331e-05,
      "loss": 0.0084,
      "step": 4295
    },
    {
      "epoch": 5.44,
      "learning_rate": 1.115732368896926e-05,
      "loss": 0.0006,
      "step": 4296
    },
    {
      "epoch": 5.44,
      "learning_rate": 1.1148282097649186e-05,
      "loss": 0.0412,
      "step": 4297
    },
    {
      "epoch": 5.44,
      "learning_rate": 1.1139240506329114e-05,
      "loss": 0.0006,
      "step": 4298
    },
    {
      "epoch": 5.44,
      "learning_rate": 1.1130198915009042e-05,
      "loss": 0.0011,
      "step": 4299
    },
    {
      "epoch": 5.44,
      "learning_rate": 1.112115732368897e-05,
      "loss": 0.0006,
      "step": 4300
    },
    {
      "epoch": 5.44,
      "learning_rate": 1.1112115732368897e-05,
      "loss": 0.0408,
      "step": 4301
    },
    {
      "epoch": 5.45,
      "learning_rate": 1.1103074141048825e-05,
      "loss": 0.0016,
      "step": 4302
    },
    {
      "epoch": 5.45,
      "learning_rate": 1.1094032549728753e-05,
      "loss": 0.0004,
      "step": 4303
    },
    {
      "epoch": 5.45,
      "learning_rate": 1.108499095840868e-05,
      "loss": 0.0005,
      "step": 4304
    },
    {
      "epoch": 5.45,
      "learning_rate": 1.1075949367088608e-05,
      "loss": 0.0009,
      "step": 4305
    },
    {
      "epoch": 5.45,
      "learning_rate": 1.1066907775768535e-05,
      "loss": 0.0008,
      "step": 4306
    },
    {
      "epoch": 5.45,
      "learning_rate": 1.1057866184448463e-05,
      "loss": 0.0121,
      "step": 4307
    },
    {
      "epoch": 5.45,
      "learning_rate": 1.1048824593128391e-05,
      "loss": 0.0006,
      "step": 4308
    },
    {
      "epoch": 5.45,
      "learning_rate": 1.1039783001808318e-05,
      "loss": 0.002,
      "step": 4309
    },
    {
      "epoch": 5.46,
      "learning_rate": 1.1030741410488246e-05,
      "loss": 0.0107,
      "step": 4310
    },
    {
      "epoch": 5.46,
      "learning_rate": 1.1021699819168174e-05,
      "loss": 0.0008,
      "step": 4311
    },
    {
      "epoch": 5.46,
      "learning_rate": 1.1012658227848103e-05,
      "loss": 0.0009,
      "step": 4312
    },
    {
      "epoch": 5.46,
      "learning_rate": 1.1003616636528029e-05,
      "loss": 0.0005,
      "step": 4313
    },
    {
      "epoch": 5.46,
      "learning_rate": 1.0994575045207957e-05,
      "loss": 0.0008,
      "step": 4314
    },
    {
      "epoch": 5.46,
      "learning_rate": 1.0985533453887886e-05,
      "loss": 0.0004,
      "step": 4315
    },
    {
      "epoch": 5.46,
      "learning_rate": 1.0976491862567812e-05,
      "loss": 0.0024,
      "step": 4316
    },
    {
      "epoch": 5.46,
      "learning_rate": 1.0967450271247739e-05,
      "loss": 0.1371,
      "step": 4317
    },
    {
      "epoch": 5.47,
      "learning_rate": 1.0958408679927667e-05,
      "loss": 0.0018,
      "step": 4318
    },
    {
      "epoch": 5.47,
      "learning_rate": 1.0949367088607595e-05,
      "loss": 0.0029,
      "step": 4319
    },
    {
      "epoch": 5.47,
      "learning_rate": 1.0940325497287523e-05,
      "loss": 0.0008,
      "step": 4320
    },
    {
      "epoch": 5.47,
      "learning_rate": 1.093128390596745e-05,
      "loss": 0.1379,
      "step": 4321
    },
    {
      "epoch": 5.47,
      "learning_rate": 1.0922242314647378e-05,
      "loss": 0.0024,
      "step": 4322
    },
    {
      "epoch": 5.47,
      "learning_rate": 1.0913200723327306e-05,
      "loss": 0.0015,
      "step": 4323
    },
    {
      "epoch": 5.47,
      "learning_rate": 1.0904159132007235e-05,
      "loss": 0.0037,
      "step": 4324
    },
    {
      "epoch": 5.47,
      "learning_rate": 1.0895117540687161e-05,
      "loss": 0.0011,
      "step": 4325
    },
    {
      "epoch": 5.48,
      "learning_rate": 1.088607594936709e-05,
      "loss": 0.0004,
      "step": 4326
    },
    {
      "epoch": 5.48,
      "learning_rate": 1.0877034358047018e-05,
      "loss": 0.001,
      "step": 4327
    },
    {
      "epoch": 5.48,
      "learning_rate": 1.0867992766726944e-05,
      "loss": 0.0035,
      "step": 4328
    },
    {
      "epoch": 5.48,
      "learning_rate": 1.085895117540687e-05,
      "loss": 0.0008,
      "step": 4329
    },
    {
      "epoch": 5.48,
      "learning_rate": 1.0849909584086799e-05,
      "loss": 0.0005,
      "step": 4330
    },
    {
      "epoch": 5.48,
      "learning_rate": 1.0840867992766727e-05,
      "loss": 0.0019,
      "step": 4331
    },
    {
      "epoch": 5.48,
      "learning_rate": 1.0831826401446655e-05,
      "loss": 0.1031,
      "step": 4332
    },
    {
      "epoch": 5.48,
      "learning_rate": 1.0822784810126582e-05,
      "loss": 0.1159,
      "step": 4333
    },
    {
      "epoch": 5.49,
      "learning_rate": 1.081374321880651e-05,
      "loss": 0.0005,
      "step": 4334
    },
    {
      "epoch": 5.49,
      "learning_rate": 1.0804701627486438e-05,
      "loss": 0.0008,
      "step": 4335
    },
    {
      "epoch": 5.49,
      "learning_rate": 1.0795660036166367e-05,
      "loss": 0.0009,
      "step": 4336
    },
    {
      "epoch": 5.49,
      "learning_rate": 1.0786618444846293e-05,
      "loss": 0.0011,
      "step": 4337
    },
    {
      "epoch": 5.49,
      "learning_rate": 1.0777576853526221e-05,
      "loss": 0.1639,
      "step": 4338
    },
    {
      "epoch": 5.49,
      "learning_rate": 1.076853526220615e-05,
      "loss": 0.004,
      "step": 4339
    },
    {
      "epoch": 5.49,
      "learning_rate": 1.0759493670886076e-05,
      "loss": 0.1192,
      "step": 4340
    },
    {
      "epoch": 5.49,
      "learning_rate": 1.0750452079566003e-05,
      "loss": 0.0004,
      "step": 4341
    },
    {
      "epoch": 5.5,
      "learning_rate": 1.0741410488245931e-05,
      "loss": 0.0008,
      "step": 4342
    },
    {
      "epoch": 5.5,
      "learning_rate": 1.073236889692586e-05,
      "loss": 0.0023,
      "step": 4343
    },
    {
      "epoch": 5.5,
      "learning_rate": 1.0723327305605788e-05,
      "loss": 0.0007,
      "step": 4344
    },
    {
      "epoch": 5.5,
      "learning_rate": 1.0714285714285714e-05,
      "loss": 0.0005,
      "step": 4345
    },
    {
      "epoch": 5.5,
      "learning_rate": 1.0705244122965642e-05,
      "loss": 0.0007,
      "step": 4346
    },
    {
      "epoch": 5.5,
      "learning_rate": 1.069620253164557e-05,
      "loss": 0.0028,
      "step": 4347
    },
    {
      "epoch": 5.5,
      "learning_rate": 1.0687160940325499e-05,
      "loss": 0.1639,
      "step": 4348
    },
    {
      "epoch": 5.5,
      "learning_rate": 1.0678119349005425e-05,
      "loss": 0.001,
      "step": 4349
    },
    {
      "epoch": 5.51,
      "learning_rate": 1.0669077757685354e-05,
      "loss": 0.0028,
      "step": 4350
    },
    {
      "epoch": 5.51,
      "learning_rate": 1.066003616636528e-05,
      "loss": 0.0006,
      "step": 4351
    },
    {
      "epoch": 5.51,
      "learning_rate": 1.0650994575045208e-05,
      "loss": 0.0048,
      "step": 4352
    },
    {
      "epoch": 5.51,
      "learning_rate": 1.0641952983725135e-05,
      "loss": 0.0057,
      "step": 4353
    },
    {
      "epoch": 5.51,
      "learning_rate": 1.0632911392405063e-05,
      "loss": 0.0004,
      "step": 4354
    },
    {
      "epoch": 5.51,
      "learning_rate": 1.0623869801084991e-05,
      "loss": 0.0033,
      "step": 4355
    },
    {
      "epoch": 5.51,
      "learning_rate": 1.061482820976492e-05,
      "loss": 0.0004,
      "step": 4356
    },
    {
      "epoch": 5.51,
      "learning_rate": 1.0605786618444846e-05,
      "loss": 0.0032,
      "step": 4357
    },
    {
      "epoch": 5.52,
      "learning_rate": 1.0596745027124774e-05,
      "loss": 0.0031,
      "step": 4358
    },
    {
      "epoch": 5.52,
      "learning_rate": 1.0587703435804703e-05,
      "loss": 0.0009,
      "step": 4359
    },
    {
      "epoch": 5.52,
      "learning_rate": 1.0578661844484631e-05,
      "loss": 0.0009,
      "step": 4360
    },
    {
      "epoch": 5.52,
      "learning_rate": 1.0569620253164557e-05,
      "loss": 0.0338,
      "step": 4361
    },
    {
      "epoch": 5.52,
      "learning_rate": 1.0560578661844486e-05,
      "loss": 0.0004,
      "step": 4362
    },
    {
      "epoch": 5.52,
      "learning_rate": 1.0551537070524412e-05,
      "loss": 0.0008,
      "step": 4363
    },
    {
      "epoch": 5.52,
      "learning_rate": 1.054249547920434e-05,
      "loss": 0.0021,
      "step": 4364
    },
    {
      "epoch": 5.52,
      "learning_rate": 1.0533453887884267e-05,
      "loss": 0.0014,
      "step": 4365
    },
    {
      "epoch": 5.53,
      "learning_rate": 1.0524412296564195e-05,
      "loss": 0.0014,
      "step": 4366
    },
    {
      "epoch": 5.53,
      "learning_rate": 1.0515370705244123e-05,
      "loss": 0.0005,
      "step": 4367
    },
    {
      "epoch": 5.53,
      "learning_rate": 1.0506329113924052e-05,
      "loss": 0.0005,
      "step": 4368
    },
    {
      "epoch": 5.53,
      "learning_rate": 1.0497287522603978e-05,
      "loss": 0.0007,
      "step": 4369
    },
    {
      "epoch": 5.53,
      "learning_rate": 1.0488245931283906e-05,
      "loss": 0.0007,
      "step": 4370
    },
    {
      "epoch": 5.53,
      "learning_rate": 1.0479204339963835e-05,
      "loss": 0.0006,
      "step": 4371
    },
    {
      "epoch": 5.53,
      "learning_rate": 1.0470162748643763e-05,
      "loss": 0.0011,
      "step": 4372
    },
    {
      "epoch": 5.54,
      "learning_rate": 1.046112115732369e-05,
      "loss": 0.0004,
      "step": 4373
    },
    {
      "epoch": 5.54,
      "learning_rate": 1.0452079566003618e-05,
      "loss": 0.1836,
      "step": 4374
    },
    {
      "epoch": 5.54,
      "learning_rate": 1.0443037974683544e-05,
      "loss": 0.0006,
      "step": 4375
    },
    {
      "epoch": 5.54,
      "learning_rate": 1.0433996383363473e-05,
      "loss": 0.0006,
      "step": 4376
    },
    {
      "epoch": 5.54,
      "learning_rate": 1.0424954792043399e-05,
      "loss": 0.0035,
      "step": 4377
    },
    {
      "epoch": 5.54,
      "learning_rate": 1.0415913200723327e-05,
      "loss": 0.0013,
      "step": 4378
    },
    {
      "epoch": 5.54,
      "learning_rate": 1.0406871609403256e-05,
      "loss": 0.0593,
      "step": 4379
    },
    {
      "epoch": 5.54,
      "learning_rate": 1.0397830018083184e-05,
      "loss": 0.0029,
      "step": 4380
    },
    {
      "epoch": 5.55,
      "learning_rate": 1.038878842676311e-05,
      "loss": 0.0012,
      "step": 4381
    },
    {
      "epoch": 5.55,
      "learning_rate": 1.0379746835443039e-05,
      "loss": 0.0011,
      "step": 4382
    },
    {
      "epoch": 5.55,
      "learning_rate": 1.0370705244122967e-05,
      "loss": 0.0004,
      "step": 4383
    },
    {
      "epoch": 5.55,
      "learning_rate": 1.0361663652802895e-05,
      "loss": 0.0039,
      "step": 4384
    },
    {
      "epoch": 5.55,
      "learning_rate": 1.0352622061482822e-05,
      "loss": 0.0007,
      "step": 4385
    },
    {
      "epoch": 5.55,
      "learning_rate": 1.0343580470162748e-05,
      "loss": 0.0006,
      "step": 4386
    },
    {
      "epoch": 5.55,
      "learning_rate": 1.0334538878842676e-05,
      "loss": 0.0009,
      "step": 4387
    },
    {
      "epoch": 5.55,
      "learning_rate": 1.0325497287522605e-05,
      "loss": 0.0006,
      "step": 4388
    },
    {
      "epoch": 5.56,
      "learning_rate": 1.0316455696202531e-05,
      "loss": 0.0019,
      "step": 4389
    },
    {
      "epoch": 5.56,
      "learning_rate": 1.030741410488246e-05,
      "loss": 0.0021,
      "step": 4390
    },
    {
      "epoch": 5.56,
      "learning_rate": 1.0298372513562388e-05,
      "loss": 0.0265,
      "step": 4391
    },
    {
      "epoch": 5.56,
      "learning_rate": 1.0289330922242316e-05,
      "loss": 0.0574,
      "step": 4392
    },
    {
      "epoch": 5.56,
      "learning_rate": 1.0280289330922242e-05,
      "loss": 0.0004,
      "step": 4393
    },
    {
      "epoch": 5.56,
      "learning_rate": 1.027124773960217e-05,
      "loss": 0.0004,
      "step": 4394
    },
    {
      "epoch": 5.56,
      "learning_rate": 1.0262206148282099e-05,
      "loss": 0.0381,
      "step": 4395
    },
    {
      "epoch": 5.56,
      "learning_rate": 1.0253164556962027e-05,
      "loss": 0.0036,
      "step": 4396
    },
    {
      "epoch": 5.57,
      "learning_rate": 1.0244122965641954e-05,
      "loss": 0.1701,
      "step": 4397
    },
    {
      "epoch": 5.57,
      "learning_rate": 1.023508137432188e-05,
      "loss": 0.01,
      "step": 4398
    },
    {
      "epoch": 5.57,
      "learning_rate": 1.0226039783001808e-05,
      "loss": 0.0015,
      "step": 4399
    },
    {
      "epoch": 5.57,
      "learning_rate": 1.0216998191681737e-05,
      "loss": 0.0035,
      "step": 4400
    },
    {
      "epoch": 5.57,
      "learning_rate": 1.0207956600361663e-05,
      "loss": 0.0005,
      "step": 4401
    },
    {
      "epoch": 5.57,
      "learning_rate": 1.0198915009041591e-05,
      "loss": 0.0009,
      "step": 4402
    },
    {
      "epoch": 5.57,
      "learning_rate": 1.018987341772152e-05,
      "loss": 0.0005,
      "step": 4403
    },
    {
      "epoch": 5.57,
      "learning_rate": 1.0180831826401448e-05,
      "loss": 0.0052,
      "step": 4404
    },
    {
      "epoch": 5.58,
      "learning_rate": 1.0171790235081374e-05,
      "loss": 0.0261,
      "step": 4405
    },
    {
      "epoch": 5.58,
      "learning_rate": 1.0162748643761303e-05,
      "loss": 0.0007,
      "step": 4406
    },
    {
      "epoch": 5.58,
      "learning_rate": 1.0153707052441231e-05,
      "loss": 0.0003,
      "step": 4407
    },
    {
      "epoch": 5.58,
      "learning_rate": 1.014466546112116e-05,
      "loss": 0.0007,
      "step": 4408
    },
    {
      "epoch": 5.58,
      "learning_rate": 1.0135623869801086e-05,
      "loss": 0.0004,
      "step": 4409
    },
    {
      "epoch": 5.58,
      "learning_rate": 1.0126582278481012e-05,
      "loss": 0.0004,
      "step": 4410
    },
    {
      "epoch": 5.58,
      "learning_rate": 1.011754068716094e-05,
      "loss": 0.0013,
      "step": 4411
    },
    {
      "epoch": 5.58,
      "learning_rate": 1.0108499095840869e-05,
      "loss": 0.0004,
      "step": 4412
    },
    {
      "epoch": 5.59,
      "learning_rate": 1.0099457504520795e-05,
      "loss": 0.11,
      "step": 4413
    },
    {
      "epoch": 5.59,
      "learning_rate": 1.0090415913200724e-05,
      "loss": 0.0004,
      "step": 4414
    },
    {
      "epoch": 5.59,
      "learning_rate": 1.0081374321880652e-05,
      "loss": 0.0004,
      "step": 4415
    },
    {
      "epoch": 5.59,
      "learning_rate": 1.007233273056058e-05,
      "loss": 0.0024,
      "step": 4416
    },
    {
      "epoch": 5.59,
      "learning_rate": 1.0063291139240507e-05,
      "loss": 0.0007,
      "step": 4417
    },
    {
      "epoch": 5.59,
      "learning_rate": 1.0054249547920435e-05,
      "loss": 0.0003,
      "step": 4418
    },
    {
      "epoch": 5.59,
      "learning_rate": 1.0045207956600363e-05,
      "loss": 0.0005,
      "step": 4419
    },
    {
      "epoch": 5.59,
      "learning_rate": 1.003616636528029e-05,
      "loss": 0.0029,
      "step": 4420
    },
    {
      "epoch": 5.6,
      "learning_rate": 1.0027124773960218e-05,
      "loss": 0.0005,
      "step": 4421
    },
    {
      "epoch": 5.6,
      "learning_rate": 1.0018083182640144e-05,
      "loss": 0.0004,
      "step": 4422
    },
    {
      "epoch": 5.6,
      "learning_rate": 1.0009041591320073e-05,
      "loss": 0.0007,
      "step": 4423
    },
    {
      "epoch": 5.6,
      "learning_rate": 1e-05,
      "loss": 0.0003,
      "step": 4424
    },
    {
      "epoch": 5.6,
      "learning_rate": 9.990958408679927e-06,
      "loss": 0.0006,
      "step": 4425
    },
    {
      "epoch": 5.6,
      "learning_rate": 9.981916817359856e-06,
      "loss": 0.0004,
      "step": 4426
    },
    {
      "epoch": 5.6,
      "learning_rate": 9.972875226039784e-06,
      "loss": 0.0012,
      "step": 4427
    },
    {
      "epoch": 5.6,
      "learning_rate": 9.963833634719712e-06,
      "loss": 0.0005,
      "step": 4428
    },
    {
      "epoch": 5.61,
      "learning_rate": 9.954792043399639e-06,
      "loss": 0.0008,
      "step": 4429
    },
    {
      "epoch": 5.61,
      "learning_rate": 9.945750452079567e-06,
      "loss": 0.0007,
      "step": 4430
    },
    {
      "epoch": 5.61,
      "learning_rate": 9.936708860759495e-06,
      "loss": 0.0016,
      "step": 4431
    },
    {
      "epoch": 5.61,
      "learning_rate": 9.927667269439422e-06,
      "loss": 0.0009,
      "step": 4432
    },
    {
      "epoch": 5.61,
      "learning_rate": 9.918625678119348e-06,
      "loss": 0.0006,
      "step": 4433
    },
    {
      "epoch": 5.61,
      "learning_rate": 9.909584086799276e-06,
      "loss": 0.1226,
      "step": 4434
    },
    {
      "epoch": 5.61,
      "learning_rate": 9.900542495479205e-06,
      "loss": 0.0096,
      "step": 4435
    },
    {
      "epoch": 5.61,
      "learning_rate": 9.891500904159133e-06,
      "loss": 0.0026,
      "step": 4436
    },
    {
      "epoch": 5.62,
      "learning_rate": 9.88245931283906e-06,
      "loss": 0.0003,
      "step": 4437
    },
    {
      "epoch": 5.62,
      "learning_rate": 9.873417721518988e-06,
      "loss": 0.0008,
      "step": 4438
    },
    {
      "epoch": 5.62,
      "learning_rate": 9.864376130198916e-06,
      "loss": 0.0015,
      "step": 4439
    },
    {
      "epoch": 5.62,
      "learning_rate": 9.855334538878844e-06,
      "loss": 0.0015,
      "step": 4440
    },
    {
      "epoch": 5.62,
      "learning_rate": 9.84629294755877e-06,
      "loss": 0.0013,
      "step": 4441
    },
    {
      "epoch": 5.62,
      "learning_rate": 9.837251356238699e-06,
      "loss": 0.0004,
      "step": 4442
    },
    {
      "epoch": 5.62,
      "learning_rate": 9.828209764918627e-06,
      "loss": 0.0004,
      "step": 4443
    },
    {
      "epoch": 5.62,
      "learning_rate": 9.819168173598554e-06,
      "loss": 0.0003,
      "step": 4444
    },
    {
      "epoch": 5.63,
      "learning_rate": 9.81012658227848e-06,
      "loss": 0.0031,
      "step": 4445
    },
    {
      "epoch": 5.63,
      "learning_rate": 9.801084990958408e-06,
      "loss": 0.0007,
      "step": 4446
    },
    {
      "epoch": 5.63,
      "learning_rate": 9.792043399638337e-06,
      "loss": 0.0006,
      "step": 4447
    },
    {
      "epoch": 5.63,
      "learning_rate": 9.783001808318265e-06,
      "loss": 0.0007,
      "step": 4448
    },
    {
      "epoch": 5.63,
      "learning_rate": 9.773960216998192e-06,
      "loss": 0.1085,
      "step": 4449
    },
    {
      "epoch": 5.63,
      "learning_rate": 9.76491862567812e-06,
      "loss": 0.0004,
      "step": 4450
    },
    {
      "epoch": 5.63,
      "learning_rate": 9.755877034358048e-06,
      "loss": 0.0005,
      "step": 4451
    },
    {
      "epoch": 5.64,
      "learning_rate": 9.746835443037976e-06,
      "loss": 0.0008,
      "step": 4452
    },
    {
      "epoch": 5.64,
      "learning_rate": 9.737793851717903e-06,
      "loss": 0.0011,
      "step": 4453
    },
    {
      "epoch": 5.64,
      "learning_rate": 9.728752260397831e-06,
      "loss": 0.0004,
      "step": 4454
    },
    {
      "epoch": 5.64,
      "learning_rate": 9.719710669077758e-06,
      "loss": 0.0004,
      "step": 4455
    },
    {
      "epoch": 5.64,
      "learning_rate": 9.710669077757686e-06,
      "loss": 0.0005,
      "step": 4456
    },
    {
      "epoch": 5.64,
      "learning_rate": 9.701627486437612e-06,
      "loss": 0.0008,
      "step": 4457
    },
    {
      "epoch": 5.64,
      "learning_rate": 9.69258589511754e-06,
      "loss": 0.0005,
      "step": 4458
    },
    {
      "epoch": 5.64,
      "learning_rate": 9.683544303797469e-06,
      "loss": 0.0004,
      "step": 4459
    },
    {
      "epoch": 5.65,
      "learning_rate": 9.674502712477397e-06,
      "loss": 0.0225,
      "step": 4460
    },
    {
      "epoch": 5.65,
      "learning_rate": 9.665461121157324e-06,
      "loss": 0.0006,
      "step": 4461
    },
    {
      "epoch": 5.65,
      "learning_rate": 9.656419529837252e-06,
      "loss": 0.0007,
      "step": 4462
    },
    {
      "epoch": 5.65,
      "learning_rate": 9.64737793851718e-06,
      "loss": 0.01,
      "step": 4463
    },
    {
      "epoch": 5.65,
      "learning_rate": 9.638336347197108e-06,
      "loss": 0.0028,
      "step": 4464
    },
    {
      "epoch": 5.65,
      "learning_rate": 9.629294755877035e-06,
      "loss": 0.027,
      "step": 4465
    },
    {
      "epoch": 5.65,
      "learning_rate": 9.620253164556963e-06,
      "loss": 0.0007,
      "step": 4466
    },
    {
      "epoch": 5.65,
      "learning_rate": 9.61121157323689e-06,
      "loss": 0.0009,
      "step": 4467
    },
    {
      "epoch": 5.66,
      "learning_rate": 9.602169981916818e-06,
      "loss": 0.0009,
      "step": 4468
    },
    {
      "epoch": 5.66,
      "learning_rate": 9.593128390596744e-06,
      "loss": 0.0004,
      "step": 4469
    },
    {
      "epoch": 5.66,
      "learning_rate": 9.584086799276673e-06,
      "loss": 0.0017,
      "step": 4470
    },
    {
      "epoch": 5.66,
      "learning_rate": 9.575045207956601e-06,
      "loss": 0.0009,
      "step": 4471
    },
    {
      "epoch": 5.66,
      "learning_rate": 9.566003616636529e-06,
      "loss": 0.0085,
      "step": 4472
    },
    {
      "epoch": 5.66,
      "learning_rate": 9.556962025316456e-06,
      "loss": 0.0005,
      "step": 4473
    },
    {
      "epoch": 5.66,
      "learning_rate": 9.547920433996384e-06,
      "loss": 0.0013,
      "step": 4474
    },
    {
      "epoch": 5.66,
      "learning_rate": 9.538878842676312e-06,
      "loss": 0.0054,
      "step": 4475
    },
    {
      "epoch": 5.67,
      "learning_rate": 9.52983725135624e-06,
      "loss": 0.0024,
      "step": 4476
    },
    {
      "epoch": 5.67,
      "learning_rate": 9.520795660036167e-06,
      "loss": 0.001,
      "step": 4477
    },
    {
      "epoch": 5.67,
      "learning_rate": 9.511754068716095e-06,
      "loss": 0.0083,
      "step": 4478
    },
    {
      "epoch": 5.67,
      "learning_rate": 9.502712477396022e-06,
      "loss": 0.0004,
      "step": 4479
    },
    {
      "epoch": 5.67,
      "learning_rate": 9.49367088607595e-06,
      "loss": 0.0004,
      "step": 4480
    },
    {
      "epoch": 5.67,
      "learning_rate": 9.484629294755876e-06,
      "loss": 0.0185,
      "step": 4481
    },
    {
      "epoch": 5.67,
      "learning_rate": 9.475587703435805e-06,
      "loss": 0.0086,
      "step": 4482
    },
    {
      "epoch": 5.67,
      "learning_rate": 9.466546112115733e-06,
      "loss": 0.0013,
      "step": 4483
    },
    {
      "epoch": 5.68,
      "learning_rate": 9.457504520795661e-06,
      "loss": 0.0004,
      "step": 4484
    },
    {
      "epoch": 5.68,
      "learning_rate": 9.448462929475588e-06,
      "loss": 0.0007,
      "step": 4485
    },
    {
      "epoch": 5.68,
      "learning_rate": 9.439421338155516e-06,
      "loss": 0.0027,
      "step": 4486
    },
    {
      "epoch": 5.68,
      "learning_rate": 9.430379746835444e-06,
      "loss": 0.0005,
      "step": 4487
    },
    {
      "epoch": 5.68,
      "learning_rate": 9.421338155515372e-06,
      "loss": 0.0003,
      "step": 4488
    },
    {
      "epoch": 5.68,
      "learning_rate": 9.412296564195299e-06,
      "loss": 0.0065,
      "step": 4489
    },
    {
      "epoch": 5.68,
      "learning_rate": 9.403254972875227e-06,
      "loss": 0.0074,
      "step": 4490
    },
    {
      "epoch": 5.68,
      "learning_rate": 9.394213381555154e-06,
      "loss": 0.0007,
      "step": 4491
    },
    {
      "epoch": 5.69,
      "learning_rate": 9.385171790235082e-06,
      "loss": 0.0151,
      "step": 4492
    },
    {
      "epoch": 5.69,
      "learning_rate": 9.376130198915009e-06,
      "loss": 0.0008,
      "step": 4493
    },
    {
      "epoch": 5.69,
      "learning_rate": 9.367088607594937e-06,
      "loss": 0.0003,
      "step": 4494
    },
    {
      "epoch": 5.69,
      "learning_rate": 9.358047016274865e-06,
      "loss": 0.0006,
      "step": 4495
    },
    {
      "epoch": 5.69,
      "learning_rate": 9.349005424954793e-06,
      "loss": 0.0003,
      "step": 4496
    },
    {
      "epoch": 5.69,
      "learning_rate": 9.33996383363472e-06,
      "loss": 0.0004,
      "step": 4497
    },
    {
      "epoch": 5.69,
      "learning_rate": 9.330922242314648e-06,
      "loss": 0.0166,
      "step": 4498
    },
    {
      "epoch": 5.69,
      "learning_rate": 9.321880650994576e-06,
      "loss": 0.0004,
      "step": 4499
    },
    {
      "epoch": 5.7,
      "learning_rate": 9.312839059674505e-06,
      "loss": 0.003,
      "step": 4500
    },
    {
      "epoch": 5.7,
      "eval_accuracy": 0.9054054054054054,
      "eval_loss": 0.47539055347442627,
      "eval_runtime": 45.0828,
      "eval_samples_per_second": 29.546,
      "eval_steps_per_second": 3.704,
      "step": 4500
    },
    {
      "epoch": 5.7,
      "learning_rate": 9.303797468354431e-06,
      "loss": 0.0005,
      "step": 4501
    },
    {
      "epoch": 5.7,
      "learning_rate": 9.294755877034358e-06,
      "loss": 0.0003,
      "step": 4502
    },
    {
      "epoch": 5.7,
      "learning_rate": 9.285714285714286e-06,
      "loss": 0.0003,
      "step": 4503
    },
    {
      "epoch": 5.7,
      "learning_rate": 9.276672694394214e-06,
      "loss": 0.0004,
      "step": 4504
    },
    {
      "epoch": 5.7,
      "learning_rate": 9.26763110307414e-06,
      "loss": 0.0005,
      "step": 4505
    },
    {
      "epoch": 5.7,
      "learning_rate": 9.258589511754069e-06,
      "loss": 0.0003,
      "step": 4506
    },
    {
      "epoch": 5.7,
      "learning_rate": 9.249547920433997e-06,
      "loss": 0.1654,
      "step": 4507
    },
    {
      "epoch": 5.71,
      "learning_rate": 9.240506329113925e-06,
      "loss": 0.0007,
      "step": 4508
    },
    {
      "epoch": 5.71,
      "learning_rate": 9.231464737793852e-06,
      "loss": 0.134,
      "step": 4509
    },
    {
      "epoch": 5.71,
      "learning_rate": 9.22242314647378e-06,
      "loss": 0.0006,
      "step": 4510
    },
    {
      "epoch": 5.71,
      "learning_rate": 9.213381555153708e-06,
      "loss": 0.0031,
      "step": 4511
    },
    {
      "epoch": 5.71,
      "learning_rate": 9.204339963833637e-06,
      "loss": 0.0005,
      "step": 4512
    },
    {
      "epoch": 5.71,
      "learning_rate": 9.195298372513563e-06,
      "loss": 0.0007,
      "step": 4513
    },
    {
      "epoch": 5.71,
      "learning_rate": 9.18625678119349e-06,
      "loss": 0.0004,
      "step": 4514
    },
    {
      "epoch": 5.71,
      "learning_rate": 9.177215189873418e-06,
      "loss": 0.0002,
      "step": 4515
    },
    {
      "epoch": 5.72,
      "learning_rate": 9.168173598553346e-06,
      "loss": 0.0037,
      "step": 4516
    },
    {
      "epoch": 5.72,
      "learning_rate": 9.159132007233273e-06,
      "loss": 0.0023,
      "step": 4517
    },
    {
      "epoch": 5.72,
      "learning_rate": 9.150090415913201e-06,
      "loss": 0.0003,
      "step": 4518
    },
    {
      "epoch": 5.72,
      "learning_rate": 9.14104882459313e-06,
      "loss": 0.0989,
      "step": 4519
    },
    {
      "epoch": 5.72,
      "learning_rate": 9.132007233273057e-06,
      "loss": 0.0003,
      "step": 4520
    },
    {
      "epoch": 5.72,
      "learning_rate": 9.122965641952984e-06,
      "loss": 0.0005,
      "step": 4521
    },
    {
      "epoch": 5.72,
      "learning_rate": 9.113924050632912e-06,
      "loss": 0.0009,
      "step": 4522
    },
    {
      "epoch": 5.72,
      "learning_rate": 9.10488245931284e-06,
      "loss": 0.0004,
      "step": 4523
    },
    {
      "epoch": 5.73,
      "learning_rate": 9.095840867992769e-06,
      "loss": 0.0009,
      "step": 4524
    },
    {
      "epoch": 5.73,
      "learning_rate": 9.086799276672695e-06,
      "loss": 0.0003,
      "step": 4525
    },
    {
      "epoch": 5.73,
      "learning_rate": 9.077757685352622e-06,
      "loss": 0.0314,
      "step": 4526
    },
    {
      "epoch": 5.73,
      "learning_rate": 9.06871609403255e-06,
      "loss": 0.0004,
      "step": 4527
    },
    {
      "epoch": 5.73,
      "learning_rate": 9.059674502712477e-06,
      "loss": 0.0003,
      "step": 4528
    },
    {
      "epoch": 5.73,
      "learning_rate": 9.050632911392405e-06,
      "loss": 0.0006,
      "step": 4529
    },
    {
      "epoch": 5.73,
      "learning_rate": 9.041591320072333e-06,
      "loss": 0.1373,
      "step": 4530
    },
    {
      "epoch": 5.73,
      "learning_rate": 9.032549728752261e-06,
      "loss": 0.0007,
      "step": 4531
    },
    {
      "epoch": 5.74,
      "learning_rate": 9.023508137432188e-06,
      "loss": 0.0027,
      "step": 4532
    },
    {
      "epoch": 5.74,
      "learning_rate": 9.014466546112116e-06,
      "loss": 0.0004,
      "step": 4533
    },
    {
      "epoch": 5.74,
      "learning_rate": 9.005424954792044e-06,
      "loss": 0.001,
      "step": 4534
    },
    {
      "epoch": 5.74,
      "learning_rate": 8.996383363471973e-06,
      "loss": 0.0017,
      "step": 4535
    },
    {
      "epoch": 5.74,
      "learning_rate": 8.987341772151899e-06,
      "loss": 0.0221,
      "step": 4536
    },
    {
      "epoch": 5.74,
      "learning_rate": 8.978300180831827e-06,
      "loss": 0.0016,
      "step": 4537
    },
    {
      "epoch": 5.74,
      "learning_rate": 8.969258589511754e-06,
      "loss": 0.0002,
      "step": 4538
    },
    {
      "epoch": 5.75,
      "learning_rate": 8.960216998191682e-06,
      "loss": 0.0004,
      "step": 4539
    },
    {
      "epoch": 5.75,
      "learning_rate": 8.951175406871609e-06,
      "loss": 0.0004,
      "step": 4540
    },
    {
      "epoch": 5.75,
      "learning_rate": 8.942133815551537e-06,
      "loss": 0.0044,
      "step": 4541
    },
    {
      "epoch": 5.75,
      "learning_rate": 8.933092224231465e-06,
      "loss": 0.0009,
      "step": 4542
    },
    {
      "epoch": 5.75,
      "learning_rate": 8.924050632911393e-06,
      "loss": 0.0004,
      "step": 4543
    },
    {
      "epoch": 5.75,
      "learning_rate": 8.91500904159132e-06,
      "loss": 0.0004,
      "step": 4544
    },
    {
      "epoch": 5.75,
      "learning_rate": 8.905967450271248e-06,
      "loss": 0.0007,
      "step": 4545
    },
    {
      "epoch": 5.75,
      "learning_rate": 8.896925858951176e-06,
      "loss": 0.0094,
      "step": 4546
    },
    {
      "epoch": 5.76,
      "learning_rate": 8.887884267631105e-06,
      "loss": 0.002,
      "step": 4547
    },
    {
      "epoch": 5.76,
      "learning_rate": 8.878842676311031e-06,
      "loss": 0.1676,
      "step": 4548
    },
    {
      "epoch": 5.76,
      "learning_rate": 8.869801084990958e-06,
      "loss": 0.0006,
      "step": 4549
    },
    {
      "epoch": 5.76,
      "learning_rate": 8.860759493670886e-06,
      "loss": 0.0237,
      "step": 4550
    },
    {
      "epoch": 5.76,
      "learning_rate": 8.851717902350814e-06,
      "loss": 0.0007,
      "step": 4551
    },
    {
      "epoch": 5.76,
      "learning_rate": 8.84267631103074e-06,
      "loss": 0.0003,
      "step": 4552
    },
    {
      "epoch": 5.76,
      "learning_rate": 8.833634719710669e-06,
      "loss": 0.0012,
      "step": 4553
    },
    {
      "epoch": 5.76,
      "learning_rate": 8.824593128390597e-06,
      "loss": 0.0003,
      "step": 4554
    },
    {
      "epoch": 5.77,
      "learning_rate": 8.815551537070525e-06,
      "loss": 0.0004,
      "step": 4555
    },
    {
      "epoch": 5.77,
      "learning_rate": 8.806509945750452e-06,
      "loss": 0.0004,
      "step": 4556
    },
    {
      "epoch": 5.77,
      "learning_rate": 8.79746835443038e-06,
      "loss": 0.0005,
      "step": 4557
    },
    {
      "epoch": 5.77,
      "learning_rate": 8.788426763110308e-06,
      "loss": 0.0009,
      "step": 4558
    },
    {
      "epoch": 5.77,
      "learning_rate": 8.779385171790237e-06,
      "loss": 0.0091,
      "step": 4559
    },
    {
      "epoch": 5.77,
      "learning_rate": 8.770343580470163e-06,
      "loss": 0.0254,
      "step": 4560
    },
    {
      "epoch": 5.77,
      "learning_rate": 8.76130198915009e-06,
      "loss": 0.0006,
      "step": 4561
    },
    {
      "epoch": 5.77,
      "learning_rate": 8.752260397830018e-06,
      "loss": 0.0004,
      "step": 4562
    },
    {
      "epoch": 5.78,
      "learning_rate": 8.743218806509946e-06,
      "loss": 0.1164,
      "step": 4563
    },
    {
      "epoch": 5.78,
      "learning_rate": 8.734177215189873e-06,
      "loss": 0.0004,
      "step": 4564
    },
    {
      "epoch": 5.78,
      "learning_rate": 8.725135623869801e-06,
      "loss": 0.2159,
      "step": 4565
    },
    {
      "epoch": 5.78,
      "learning_rate": 8.71609403254973e-06,
      "loss": 0.0004,
      "step": 4566
    },
    {
      "epoch": 5.78,
      "learning_rate": 8.707052441229658e-06,
      "loss": 0.0155,
      "step": 4567
    },
    {
      "epoch": 5.78,
      "learning_rate": 8.698010849909584e-06,
      "loss": 0.0009,
      "step": 4568
    },
    {
      "epoch": 5.78,
      "learning_rate": 8.688969258589512e-06,
      "loss": 0.0014,
      "step": 4569
    },
    {
      "epoch": 5.78,
      "learning_rate": 8.67992766726944e-06,
      "loss": 0.0003,
      "step": 4570
    },
    {
      "epoch": 5.79,
      "learning_rate": 8.670886075949367e-06,
      "loss": 0.0004,
      "step": 4571
    },
    {
      "epoch": 5.79,
      "learning_rate": 8.661844484629295e-06,
      "loss": 0.0155,
      "step": 4572
    },
    {
      "epoch": 5.79,
      "learning_rate": 8.652802893309222e-06,
      "loss": 0.0005,
      "step": 4573
    },
    {
      "epoch": 5.79,
      "learning_rate": 8.64376130198915e-06,
      "loss": 0.0006,
      "step": 4574
    },
    {
      "epoch": 5.79,
      "learning_rate": 8.634719710669078e-06,
      "loss": 0.0062,
      "step": 4575
    },
    {
      "epoch": 5.79,
      "learning_rate": 8.625678119349005e-06,
      "loss": 0.0028,
      "step": 4576
    },
    {
      "epoch": 5.79,
      "learning_rate": 8.616636528028933e-06,
      "loss": 0.001,
      "step": 4577
    },
    {
      "epoch": 5.79,
      "learning_rate": 8.607594936708861e-06,
      "loss": 0.001,
      "step": 4578
    },
    {
      "epoch": 5.8,
      "learning_rate": 8.59855334538879e-06,
      "loss": 0.2511,
      "step": 4579
    },
    {
      "epoch": 5.8,
      "learning_rate": 8.589511754068716e-06,
      "loss": 0.029,
      "step": 4580
    },
    {
      "epoch": 5.8,
      "learning_rate": 8.580470162748644e-06,
      "loss": 0.0008,
      "step": 4581
    },
    {
      "epoch": 5.8,
      "learning_rate": 8.571428571428573e-06,
      "loss": 0.0441,
      "step": 4582
    },
    {
      "epoch": 5.8,
      "learning_rate": 8.5623869801085e-06,
      "loss": 0.0998,
      "step": 4583
    },
    {
      "epoch": 5.8,
      "learning_rate": 8.553345388788427e-06,
      "loss": 0.0012,
      "step": 4584
    },
    {
      "epoch": 5.8,
      "learning_rate": 8.544303797468354e-06,
      "loss": 0.0069,
      "step": 4585
    },
    {
      "epoch": 5.8,
      "learning_rate": 8.535262206148282e-06,
      "loss": 0.0007,
      "step": 4586
    },
    {
      "epoch": 5.81,
      "learning_rate": 8.52622061482821e-06,
      "loss": 0.0006,
      "step": 4587
    },
    {
      "epoch": 5.81,
      "learning_rate": 8.517179023508137e-06,
      "loss": 0.0006,
      "step": 4588
    },
    {
      "epoch": 5.81,
      "learning_rate": 8.508137432188065e-06,
      "loss": 0.0021,
      "step": 4589
    },
    {
      "epoch": 5.81,
      "learning_rate": 8.499095840867993e-06,
      "loss": 0.0007,
      "step": 4590
    },
    {
      "epoch": 5.81,
      "learning_rate": 8.490054249547922e-06,
      "loss": 0.0019,
      "step": 4591
    },
    {
      "epoch": 5.81,
      "learning_rate": 8.481012658227848e-06,
      "loss": 0.0008,
      "step": 4592
    },
    {
      "epoch": 5.81,
      "learning_rate": 8.471971066907776e-06,
      "loss": 0.0015,
      "step": 4593
    },
    {
      "epoch": 5.81,
      "learning_rate": 8.462929475587705e-06,
      "loss": 0.0019,
      "step": 4594
    },
    {
      "epoch": 5.82,
      "learning_rate": 8.453887884267631e-06,
      "loss": 0.0004,
      "step": 4595
    },
    {
      "epoch": 5.82,
      "learning_rate": 8.444846292947558e-06,
      "loss": 0.0009,
      "step": 4596
    },
    {
      "epoch": 5.82,
      "learning_rate": 8.435804701627486e-06,
      "loss": 0.0013,
      "step": 4597
    },
    {
      "epoch": 5.82,
      "learning_rate": 8.426763110307414e-06,
      "loss": 0.0034,
      "step": 4598
    },
    {
      "epoch": 5.82,
      "learning_rate": 8.417721518987342e-06,
      "loss": 0.0498,
      "step": 4599
    },
    {
      "epoch": 5.82,
      "learning_rate": 8.408679927667269e-06,
      "loss": 0.0005,
      "step": 4600
    },
    {
      "epoch": 5.82,
      "learning_rate": 8.399638336347197e-06,
      "loss": 0.0006,
      "step": 4601
    },
    {
      "epoch": 5.82,
      "learning_rate": 8.390596745027126e-06,
      "loss": 0.0004,
      "step": 4602
    },
    {
      "epoch": 5.83,
      "learning_rate": 8.381555153707054e-06,
      "loss": 0.0007,
      "step": 4603
    },
    {
      "epoch": 5.83,
      "learning_rate": 8.37251356238698e-06,
      "loss": 0.0284,
      "step": 4604
    },
    {
      "epoch": 5.83,
      "learning_rate": 8.363471971066909e-06,
      "loss": 0.0127,
      "step": 4605
    },
    {
      "epoch": 5.83,
      "learning_rate": 8.354430379746837e-06,
      "loss": 0.0007,
      "step": 4606
    },
    {
      "epoch": 5.83,
      "learning_rate": 8.345388788426763e-06,
      "loss": 0.0008,
      "step": 4607
    },
    {
      "epoch": 5.83,
      "learning_rate": 8.33634719710669e-06,
      "loss": 0.0009,
      "step": 4608
    },
    {
      "epoch": 5.83,
      "learning_rate": 8.327305605786618e-06,
      "loss": 0.0004,
      "step": 4609
    },
    {
      "epoch": 5.83,
      "learning_rate": 8.318264014466546e-06,
      "loss": 0.0706,
      "step": 4610
    },
    {
      "epoch": 5.84,
      "learning_rate": 8.309222423146475e-06,
      "loss": 0.0012,
      "step": 4611
    },
    {
      "epoch": 5.84,
      "learning_rate": 8.300180831826401e-06,
      "loss": 0.0589,
      "step": 4612
    },
    {
      "epoch": 5.84,
      "learning_rate": 8.29113924050633e-06,
      "loss": 0.0006,
      "step": 4613
    },
    {
      "epoch": 5.84,
      "learning_rate": 8.282097649186258e-06,
      "loss": 0.0006,
      "step": 4614
    },
    {
      "epoch": 5.84,
      "learning_rate": 8.273056057866186e-06,
      "loss": 0.0029,
      "step": 4615
    },
    {
      "epoch": 5.84,
      "learning_rate": 8.264014466546112e-06,
      "loss": 0.113,
      "step": 4616
    },
    {
      "epoch": 5.84,
      "learning_rate": 8.25497287522604e-06,
      "loss": 0.0057,
      "step": 4617
    },
    {
      "epoch": 5.85,
      "learning_rate": 8.245931283905967e-06,
      "loss": 0.0004,
      "step": 4618
    },
    {
      "epoch": 5.85,
      "learning_rate": 8.236889692585895e-06,
      "loss": 0.0004,
      "step": 4619
    },
    {
      "epoch": 5.85,
      "learning_rate": 8.227848101265822e-06,
      "loss": 0.0003,
      "step": 4620
    },
    {
      "epoch": 5.85,
      "learning_rate": 8.21880650994575e-06,
      "loss": 0.0009,
      "step": 4621
    },
    {
      "epoch": 5.85,
      "learning_rate": 8.209764918625678e-06,
      "loss": 0.0003,
      "step": 4622
    },
    {
      "epoch": 5.85,
      "learning_rate": 8.200723327305607e-06,
      "loss": 0.0003,
      "step": 4623
    },
    {
      "epoch": 5.85,
      "learning_rate": 8.191681735985533e-06,
      "loss": 0.0006,
      "step": 4624
    },
    {
      "epoch": 5.85,
      "learning_rate": 8.182640144665461e-06,
      "loss": 0.0011,
      "step": 4625
    },
    {
      "epoch": 5.86,
      "learning_rate": 8.17359855334539e-06,
      "loss": 0.0016,
      "step": 4626
    },
    {
      "epoch": 5.86,
      "learning_rate": 8.164556962025318e-06,
      "loss": 0.0007,
      "step": 4627
    },
    {
      "epoch": 5.86,
      "learning_rate": 8.155515370705244e-06,
      "loss": 0.0018,
      "step": 4628
    },
    {
      "epoch": 5.86,
      "learning_rate": 8.146473779385173e-06,
      "loss": 0.0003,
      "step": 4629
    },
    {
      "epoch": 5.86,
      "learning_rate": 8.1374321880651e-06,
      "loss": 0.0004,
      "step": 4630
    },
    {
      "epoch": 5.86,
      "learning_rate": 8.128390596745027e-06,
      "loss": 0.0008,
      "step": 4631
    },
    {
      "epoch": 5.86,
      "learning_rate": 8.119349005424954e-06,
      "loss": 0.0007,
      "step": 4632
    },
    {
      "epoch": 5.86,
      "learning_rate": 8.110307414104882e-06,
      "loss": 0.0003,
      "step": 4633
    },
    {
      "epoch": 5.87,
      "learning_rate": 8.10126582278481e-06,
      "loss": 0.0004,
      "step": 4634
    },
    {
      "epoch": 5.87,
      "learning_rate": 8.092224231464739e-06,
      "loss": 0.0004,
      "step": 4635
    },
    {
      "epoch": 5.87,
      "learning_rate": 8.083182640144665e-06,
      "loss": 0.0002,
      "step": 4636
    },
    {
      "epoch": 5.87,
      "learning_rate": 8.074141048824594e-06,
      "loss": 0.0012,
      "step": 4637
    },
    {
      "epoch": 5.87,
      "learning_rate": 8.065099457504522e-06,
      "loss": 0.0002,
      "step": 4638
    },
    {
      "epoch": 5.87,
      "learning_rate": 8.05605786618445e-06,
      "loss": 0.0007,
      "step": 4639
    },
    {
      "epoch": 5.87,
      "learning_rate": 8.047016274864377e-06,
      "loss": 0.0002,
      "step": 4640
    },
    {
      "epoch": 5.87,
      "learning_rate": 8.037974683544305e-06,
      "loss": 0.0005,
      "step": 4641
    },
    {
      "epoch": 5.88,
      "learning_rate": 8.028933092224231e-06,
      "loss": 0.0009,
      "step": 4642
    },
    {
      "epoch": 5.88,
      "learning_rate": 8.01989150090416e-06,
      "loss": 0.0686,
      "step": 4643
    },
    {
      "epoch": 5.88,
      "learning_rate": 8.010849909584086e-06,
      "loss": 0.0006,
      "step": 4644
    },
    {
      "epoch": 5.88,
      "learning_rate": 8.001808318264014e-06,
      "loss": 0.0003,
      "step": 4645
    },
    {
      "epoch": 5.88,
      "learning_rate": 7.992766726943943e-06,
      "loss": 0.0022,
      "step": 4646
    },
    {
      "epoch": 5.88,
      "learning_rate": 7.98372513562387e-06,
      "loss": 0.0021,
      "step": 4647
    },
    {
      "epoch": 5.88,
      "learning_rate": 7.974683544303797e-06,
      "loss": 0.0047,
      "step": 4648
    },
    {
      "epoch": 5.88,
      "learning_rate": 7.965641952983726e-06,
      "loss": 0.0021,
      "step": 4649
    },
    {
      "epoch": 5.89,
      "learning_rate": 7.956600361663654e-06,
      "loss": 0.0701,
      "step": 4650
    },
    {
      "epoch": 5.89,
      "learning_rate": 7.947558770343582e-06,
      "loss": 0.0006,
      "step": 4651
    },
    {
      "epoch": 5.89,
      "learning_rate": 7.938517179023509e-06,
      "loss": 0.0126,
      "step": 4652
    },
    {
      "epoch": 5.89,
      "learning_rate": 7.929475587703437e-06,
      "loss": 0.0009,
      "step": 4653
    },
    {
      "epoch": 5.89,
      "learning_rate": 7.920433996383363e-06,
      "loss": 0.1196,
      "step": 4654
    },
    {
      "epoch": 5.89,
      "learning_rate": 7.911392405063292e-06,
      "loss": 0.0497,
      "step": 4655
    },
    {
      "epoch": 5.89,
      "learning_rate": 7.902350813743218e-06,
      "loss": 0.0005,
      "step": 4656
    },
    {
      "epoch": 5.89,
      "learning_rate": 7.893309222423146e-06,
      "loss": 0.0005,
      "step": 4657
    },
    {
      "epoch": 5.9,
      "learning_rate": 7.884267631103075e-06,
      "loss": 0.0007,
      "step": 4658
    },
    {
      "epoch": 5.9,
      "learning_rate": 7.875226039783003e-06,
      "loss": 0.0002,
      "step": 4659
    },
    {
      "epoch": 5.9,
      "learning_rate": 7.86618444846293e-06,
      "loss": 0.0003,
      "step": 4660
    },
    {
      "epoch": 5.9,
      "learning_rate": 7.857142857142858e-06,
      "loss": 0.2239,
      "step": 4661
    },
    {
      "epoch": 5.9,
      "learning_rate": 7.848101265822786e-06,
      "loss": 0.0003,
      "step": 4662
    },
    {
      "epoch": 5.9,
      "learning_rate": 7.839059674502714e-06,
      "loss": 0.0006,
      "step": 4663
    },
    {
      "epoch": 5.9,
      "learning_rate": 7.83001808318264e-06,
      "loss": 0.0003,
      "step": 4664
    },
    {
      "epoch": 5.9,
      "learning_rate": 7.820976491862567e-06,
      "loss": 0.0032,
      "step": 4665
    },
    {
      "epoch": 5.91,
      "learning_rate": 7.811934900542495e-06,
      "loss": 0.0005,
      "step": 4666
    },
    {
      "epoch": 5.91,
      "learning_rate": 7.802893309222424e-06,
      "loss": 0.0005,
      "step": 4667
    },
    {
      "epoch": 5.91,
      "learning_rate": 7.79385171790235e-06,
      "loss": 0.0005,
      "step": 4668
    },
    {
      "epoch": 5.91,
      "learning_rate": 7.784810126582278e-06,
      "loss": 0.0014,
      "step": 4669
    },
    {
      "epoch": 5.91,
      "learning_rate": 7.775768535262207e-06,
      "loss": 0.0004,
      "step": 4670
    },
    {
      "epoch": 5.91,
      "learning_rate": 7.766726943942135e-06,
      "loss": 0.0004,
      "step": 4671
    },
    {
      "epoch": 5.91,
      "learning_rate": 7.757685352622062e-06,
      "loss": 0.0003,
      "step": 4672
    },
    {
      "epoch": 5.91,
      "learning_rate": 7.74864376130199e-06,
      "loss": 0.0005,
      "step": 4673
    },
    {
      "epoch": 5.92,
      "learning_rate": 7.739602169981918e-06,
      "loss": 0.0004,
      "step": 4674
    },
    {
      "epoch": 5.92,
      "learning_rate": 7.730560578661846e-06,
      "loss": 0.001,
      "step": 4675
    },
    {
      "epoch": 5.92,
      "learning_rate": 7.721518987341773e-06,
      "loss": 0.0038,
      "step": 4676
    },
    {
      "epoch": 5.92,
      "learning_rate": 7.7124773960217e-06,
      "loss": 0.0007,
      "step": 4677
    },
    {
      "epoch": 5.92,
      "learning_rate": 7.703435804701628e-06,
      "loss": 0.0003,
      "step": 4678
    },
    {
      "epoch": 5.92,
      "learning_rate": 7.694394213381556e-06,
      "loss": 0.0003,
      "step": 4679
    },
    {
      "epoch": 5.92,
      "learning_rate": 7.685352622061482e-06,
      "loss": 0.1132,
      "step": 4680
    },
    {
      "epoch": 5.92,
      "learning_rate": 7.67631103074141e-06,
      "loss": 0.2606,
      "step": 4681
    },
    {
      "epoch": 5.93,
      "learning_rate": 7.667269439421339e-06,
      "loss": 0.002,
      "step": 4682
    },
    {
      "epoch": 5.93,
      "learning_rate": 7.658227848101267e-06,
      "loss": 0.001,
      "step": 4683
    },
    {
      "epoch": 5.93,
      "learning_rate": 7.649186256781194e-06,
      "loss": 0.0002,
      "step": 4684
    },
    {
      "epoch": 5.93,
      "learning_rate": 7.640144665461122e-06,
      "loss": 0.054,
      "step": 4685
    },
    {
      "epoch": 5.93,
      "learning_rate": 7.63110307414105e-06,
      "loss": 0.0003,
      "step": 4686
    },
    {
      "epoch": 5.93,
      "learning_rate": 7.6220614828209774e-06,
      "loss": 0.0007,
      "step": 4687
    },
    {
      "epoch": 5.93,
      "learning_rate": 7.613019891500904e-06,
      "loss": 0.001,
      "step": 4688
    },
    {
      "epoch": 5.93,
      "learning_rate": 7.603978300180832e-06,
      "loss": 0.0005,
      "step": 4689
    },
    {
      "epoch": 5.94,
      "learning_rate": 7.5949367088607605e-06,
      "loss": 0.0014,
      "step": 4690
    },
    {
      "epoch": 5.94,
      "learning_rate": 7.585895117540688e-06,
      "loss": 0.0009,
      "step": 4691
    },
    {
      "epoch": 5.94,
      "learning_rate": 7.576853526220614e-06,
      "loss": 0.0493,
      "step": 4692
    },
    {
      "epoch": 5.94,
      "learning_rate": 7.567811934900543e-06,
      "loss": 0.0006,
      "step": 4693
    },
    {
      "epoch": 5.94,
      "learning_rate": 7.558770343580471e-06,
      "loss": 0.0017,
      "step": 4694
    },
    {
      "epoch": 5.94,
      "learning_rate": 7.549728752260399e-06,
      "loss": 0.0017,
      "step": 4695
    },
    {
      "epoch": 5.94,
      "learning_rate": 7.540687160940326e-06,
      "loss": 0.0014,
      "step": 4696
    },
    {
      "epoch": 5.94,
      "learning_rate": 7.531645569620253e-06,
      "loss": 0.0007,
      "step": 4697
    },
    {
      "epoch": 5.95,
      "learning_rate": 7.522603978300181e-06,
      "loss": 0.0045,
      "step": 4698
    },
    {
      "epoch": 5.95,
      "learning_rate": 7.5135623869801095e-06,
      "loss": 0.0004,
      "step": 4699
    },
    {
      "epoch": 5.95,
      "learning_rate": 7.504520795660036e-06,
      "loss": 0.225,
      "step": 4700
    },
    {
      "epoch": 5.95,
      "learning_rate": 7.495479204339964e-06,
      "loss": 0.0011,
      "step": 4701
    },
    {
      "epoch": 5.95,
      "learning_rate": 7.486437613019892e-06,
      "loss": 0.0011,
      "step": 4702
    },
    {
      "epoch": 5.95,
      "learning_rate": 7.47739602169982e-06,
      "loss": 0.0006,
      "step": 4703
    },
    {
      "epoch": 5.95,
      "learning_rate": 7.4683544303797465e-06,
      "loss": 0.0053,
      "step": 4704
    },
    {
      "epoch": 5.96,
      "learning_rate": 7.459312839059675e-06,
      "loss": 0.0006,
      "step": 4705
    },
    {
      "epoch": 5.96,
      "learning_rate": 7.450271247739603e-06,
      "loss": 0.078,
      "step": 4706
    },
    {
      "epoch": 5.96,
      "learning_rate": 7.441229656419531e-06,
      "loss": 0.0004,
      "step": 4707
    },
    {
      "epoch": 5.96,
      "learning_rate": 7.432188065099458e-06,
      "loss": 0.0003,
      "step": 4708
    },
    {
      "epoch": 5.96,
      "learning_rate": 7.423146473779385e-06,
      "loss": 0.001,
      "step": 4709
    },
    {
      "epoch": 5.96,
      "learning_rate": 7.414104882459313e-06,
      "loss": 0.0005,
      "step": 4710
    },
    {
      "epoch": 5.96,
      "learning_rate": 7.405063291139242e-06,
      "loss": 0.0008,
      "step": 4711
    },
    {
      "epoch": 5.96,
      "learning_rate": 7.396021699819168e-06,
      "loss": 0.0046,
      "step": 4712
    },
    {
      "epoch": 5.97,
      "learning_rate": 7.386980108499096e-06,
      "loss": 0.0005,
      "step": 4713
    },
    {
      "epoch": 5.97,
      "learning_rate": 7.377938517179024e-06,
      "loss": 0.0005,
      "step": 4714
    },
    {
      "epoch": 5.97,
      "learning_rate": 7.368896925858952e-06,
      "loss": 0.0007,
      "step": 4715
    },
    {
      "epoch": 5.97,
      "learning_rate": 7.3598553345388786e-06,
      "loss": 0.0009,
      "step": 4716
    },
    {
      "epoch": 5.97,
      "learning_rate": 7.350813743218807e-06,
      "loss": 0.0004,
      "step": 4717
    },
    {
      "epoch": 5.97,
      "learning_rate": 7.341772151898735e-06,
      "loss": 0.0011,
      "step": 4718
    },
    {
      "epoch": 5.97,
      "learning_rate": 7.3327305605786624e-06,
      "loss": 0.0005,
      "step": 4719
    },
    {
      "epoch": 5.97,
      "learning_rate": 7.32368896925859e-06,
      "loss": 0.0005,
      "step": 4720
    },
    {
      "epoch": 5.98,
      "learning_rate": 7.314647377938517e-06,
      "loss": 0.0004,
      "step": 4721
    },
    {
      "epoch": 5.98,
      "learning_rate": 7.3056057866184454e-06,
      "loss": 0.0005,
      "step": 4722
    },
    {
      "epoch": 5.98,
      "learning_rate": 7.296564195298374e-06,
      "loss": 0.0008,
      "step": 4723
    },
    {
      "epoch": 5.98,
      "learning_rate": 7.2875226039783e-06,
      "loss": 0.0015,
      "step": 4724
    },
    {
      "epoch": 5.98,
      "learning_rate": 7.2784810126582285e-06,
      "loss": 0.016,
      "step": 4725
    },
    {
      "epoch": 5.98,
      "learning_rate": 7.269439421338156e-06,
      "loss": 0.0012,
      "step": 4726
    },
    {
      "epoch": 5.98,
      "learning_rate": 7.260397830018084e-06,
      "loss": 0.0008,
      "step": 4727
    },
    {
      "epoch": 5.98,
      "learning_rate": 7.251356238698011e-06,
      "loss": 0.0005,
      "step": 4728
    },
    {
      "epoch": 5.99,
      "learning_rate": 7.242314647377939e-06,
      "loss": 0.0003,
      "step": 4729
    },
    {
      "epoch": 5.99,
      "learning_rate": 7.233273056057867e-06,
      "loss": 0.0008,
      "step": 4730
    },
    {
      "epoch": 5.99,
      "learning_rate": 7.2242314647377945e-06,
      "loss": 0.0005,
      "step": 4731
    },
    {
      "epoch": 5.99,
      "learning_rate": 7.215189873417721e-06,
      "loss": 0.0015,
      "step": 4732
    },
    {
      "epoch": 5.99,
      "learning_rate": 7.206148282097649e-06,
      "loss": 0.0005,
      "step": 4733
    },
    {
      "epoch": 5.99,
      "learning_rate": 7.1971066907775775e-06,
      "loss": 0.0006,
      "step": 4734
    },
    {
      "epoch": 5.99,
      "learning_rate": 7.188065099457506e-06,
      "loss": 0.0004,
      "step": 4735
    },
    {
      "epoch": 5.99,
      "learning_rate": 7.179023508137432e-06,
      "loss": 0.0009,
      "step": 4736
    },
    {
      "epoch": 6.0,
      "learning_rate": 7.1699819168173605e-06,
      "loss": 0.0007,
      "step": 4737
    },
    {
      "epoch": 6.0,
      "learning_rate": 7.160940325497288e-06,
      "loss": 0.062,
      "step": 4738
    },
    {
      "epoch": 6.0,
      "learning_rate": 7.151898734177216e-06,
      "loss": 0.0004,
      "step": 4739
    },
    {
      "epoch": 6.0,
      "learning_rate": 7.142857142857143e-06,
      "loss": 0.0011,
      "step": 4740
    },
    {
      "epoch": 6.0,
      "learning_rate": 7.133815551537071e-06,
      "loss": 0.0017,
      "step": 4741
    },
    {
      "epoch": 6.0,
      "learning_rate": 7.124773960216999e-06,
      "loss": 0.0211,
      "step": 4742
    },
    {
      "epoch": 6.0,
      "learning_rate": 7.115732368896927e-06,
      "loss": 0.0004,
      "step": 4743
    },
    {
      "epoch": 6.01,
      "learning_rate": 7.106690777576853e-06,
      "loss": 0.0019,
      "step": 4744
    },
    {
      "epoch": 6.01,
      "learning_rate": 7.097649186256781e-06,
      "loss": 0.0003,
      "step": 4745
    },
    {
      "epoch": 6.01,
      "learning_rate": 7.08860759493671e-06,
      "loss": 0.0007,
      "step": 4746
    },
    {
      "epoch": 6.01,
      "learning_rate": 7.079566003616638e-06,
      "loss": 0.0003,
      "step": 4747
    },
    {
      "epoch": 6.01,
      "learning_rate": 7.070524412296564e-06,
      "loss": 0.0004,
      "step": 4748
    },
    {
      "epoch": 6.01,
      "learning_rate": 7.061482820976492e-06,
      "loss": 0.0022,
      "step": 4749
    },
    {
      "epoch": 6.01,
      "learning_rate": 7.05244122965642e-06,
      "loss": 0.0004,
      "step": 4750
    },
    {
      "epoch": 6.01,
      "eval_accuracy": 0.9061561561561562,
      "eval_loss": 0.4733130633831024,
      "eval_runtime": 45.1372,
      "eval_samples_per_second": 29.51,
      "eval_steps_per_second": 3.7,
      "step": 4750
    },
    {
      "epoch": 6.01,
      "learning_rate": 7.043399638336348e-06,
      "loss": 0.0005,
      "step": 4751
    },
    {
      "epoch": 6.02,
      "learning_rate": 7.034358047016275e-06,
      "loss": 0.0061,
      "step": 4752
    },
    {
      "epoch": 6.02,
      "learning_rate": 7.025316455696203e-06,
      "loss": 0.0004,
      "step": 4753
    },
    {
      "epoch": 6.02,
      "learning_rate": 7.016274864376131e-06,
      "loss": 0.0116,
      "step": 4754
    },
    {
      "epoch": 6.02,
      "learning_rate": 7.007233273056059e-06,
      "loss": 0.0004,
      "step": 4755
    },
    {
      "epoch": 6.02,
      "learning_rate": 6.998191681735985e-06,
      "loss": 0.0017,
      "step": 4756
    },
    {
      "epoch": 6.02,
      "learning_rate": 6.9891500904159134e-06,
      "loss": 0.0005,
      "step": 4757
    },
    {
      "epoch": 6.02,
      "learning_rate": 6.980108499095842e-06,
      "loss": 0.0003,
      "step": 4758
    },
    {
      "epoch": 6.02,
      "learning_rate": 6.97106690777577e-06,
      "loss": 0.0004,
      "step": 4759
    },
    {
      "epoch": 6.03,
      "learning_rate": 6.9620253164556965e-06,
      "loss": 0.0005,
      "step": 4760
    },
    {
      "epoch": 6.03,
      "learning_rate": 6.952983725135624e-06,
      "loss": 0.0005,
      "step": 4761
    },
    {
      "epoch": 6.03,
      "learning_rate": 6.943942133815552e-06,
      "loss": 0.0005,
      "step": 4762
    },
    {
      "epoch": 6.03,
      "learning_rate": 6.93490054249548e-06,
      "loss": 0.0044,
      "step": 4763
    },
    {
      "epoch": 6.03,
      "learning_rate": 6.925858951175407e-06,
      "loss": 0.0006,
      "step": 4764
    },
    {
      "epoch": 6.03,
      "learning_rate": 6.916817359855335e-06,
      "loss": 0.0033,
      "step": 4765
    },
    {
      "epoch": 6.03,
      "learning_rate": 6.9077757685352625e-06,
      "loss": 0.0007,
      "step": 4766
    },
    {
      "epoch": 6.03,
      "learning_rate": 6.898734177215191e-06,
      "loss": 0.0003,
      "step": 4767
    },
    {
      "epoch": 6.04,
      "learning_rate": 6.889692585895117e-06,
      "loss": 0.0004,
      "step": 4768
    },
    {
      "epoch": 6.04,
      "learning_rate": 6.8806509945750455e-06,
      "loss": 0.0003,
      "step": 4769
    },
    {
      "epoch": 6.04,
      "learning_rate": 6.871609403254974e-06,
      "loss": 0.0005,
      "step": 4770
    },
    {
      "epoch": 6.04,
      "learning_rate": 6.862567811934901e-06,
      "loss": 0.0004,
      "step": 4771
    },
    {
      "epoch": 6.04,
      "learning_rate": 6.8535262206148285e-06,
      "loss": 0.0005,
      "step": 4772
    },
    {
      "epoch": 6.04,
      "learning_rate": 6.844484629294756e-06,
      "loss": 0.0956,
      "step": 4773
    },
    {
      "epoch": 6.04,
      "learning_rate": 6.835443037974684e-06,
      "loss": 0.0002,
      "step": 4774
    },
    {
      "epoch": 6.04,
      "learning_rate": 6.826401446654612e-06,
      "loss": 0.0003,
      "step": 4775
    },
    {
      "epoch": 6.05,
      "learning_rate": 6.817359855334539e-06,
      "loss": 0.0004,
      "step": 4776
    },
    {
      "epoch": 6.05,
      "learning_rate": 6.808318264014467e-06,
      "loss": 0.0005,
      "step": 4777
    },
    {
      "epoch": 6.05,
      "learning_rate": 6.799276672694395e-06,
      "loss": 0.0005,
      "step": 4778
    },
    {
      "epoch": 6.05,
      "learning_rate": 6.790235081374321e-06,
      "loss": 0.0006,
      "step": 4779
    },
    {
      "epoch": 6.05,
      "learning_rate": 6.781193490054249e-06,
      "loss": 0.0097,
      "step": 4780
    },
    {
      "epoch": 6.05,
      "learning_rate": 6.772151898734178e-06,
      "loss": 0.0003,
      "step": 4781
    },
    {
      "epoch": 6.05,
      "learning_rate": 6.763110307414106e-06,
      "loss": 0.0002,
      "step": 4782
    },
    {
      "epoch": 6.05,
      "learning_rate": 6.754068716094032e-06,
      "loss": 0.0026,
      "step": 4783
    },
    {
      "epoch": 6.06,
      "learning_rate": 6.745027124773961e-06,
      "loss": 0.0007,
      "step": 4784
    },
    {
      "epoch": 6.06,
      "learning_rate": 6.735985533453888e-06,
      "loss": 0.0019,
      "step": 4785
    },
    {
      "epoch": 6.06,
      "learning_rate": 6.726943942133816e-06,
      "loss": 0.0005,
      "step": 4786
    },
    {
      "epoch": 6.06,
      "learning_rate": 6.717902350813743e-06,
      "loss": 0.0005,
      "step": 4787
    },
    {
      "epoch": 6.06,
      "learning_rate": 6.708860759493671e-06,
      "loss": 0.0003,
      "step": 4788
    },
    {
      "epoch": 6.06,
      "learning_rate": 6.699819168173599e-06,
      "loss": 0.0009,
      "step": 4789
    },
    {
      "epoch": 6.06,
      "learning_rate": 6.690777576853527e-06,
      "loss": 0.0026,
      "step": 4790
    },
    {
      "epoch": 6.06,
      "learning_rate": 6.681735985533453e-06,
      "loss": 0.0004,
      "step": 4791
    },
    {
      "epoch": 6.07,
      "learning_rate": 6.6726943942133814e-06,
      "loss": 0.0015,
      "step": 4792
    },
    {
      "epoch": 6.07,
      "learning_rate": 6.66365280289331e-06,
      "loss": 0.0003,
      "step": 4793
    },
    {
      "epoch": 6.07,
      "learning_rate": 6.654611211573238e-06,
      "loss": 0.0004,
      "step": 4794
    },
    {
      "epoch": 6.07,
      "learning_rate": 6.6455696202531645e-06,
      "loss": 0.0008,
      "step": 4795
    },
    {
      "epoch": 6.07,
      "learning_rate": 6.636528028933092e-06,
      "loss": 0.0004,
      "step": 4796
    },
    {
      "epoch": 6.07,
      "learning_rate": 6.62748643761302e-06,
      "loss": 0.0005,
      "step": 4797
    },
    {
      "epoch": 6.07,
      "learning_rate": 6.618444846292948e-06,
      "loss": 0.0007,
      "step": 4798
    },
    {
      "epoch": 6.07,
      "learning_rate": 6.609403254972875e-06,
      "loss": 0.0003,
      "step": 4799
    },
    {
      "epoch": 6.08,
      "learning_rate": 6.600361663652803e-06,
      "loss": 0.0003,
      "step": 4800
    },
    {
      "epoch": 6.08,
      "learning_rate": 6.5913200723327305e-06,
      "loss": 0.0003,
      "step": 4801
    },
    {
      "epoch": 6.08,
      "learning_rate": 6.582278481012659e-06,
      "loss": 0.0008,
      "step": 4802
    },
    {
      "epoch": 6.08,
      "learning_rate": 6.573236889692585e-06,
      "loss": 0.0004,
      "step": 4803
    },
    {
      "epoch": 6.08,
      "learning_rate": 6.5641952983725135e-06,
      "loss": 0.0004,
      "step": 4804
    },
    {
      "epoch": 6.08,
      "learning_rate": 6.555153707052442e-06,
      "loss": 0.0004,
      "step": 4805
    },
    {
      "epoch": 6.08,
      "learning_rate": 6.54611211573237e-06,
      "loss": 0.0002,
      "step": 4806
    },
    {
      "epoch": 6.08,
      "learning_rate": 6.5370705244122965e-06,
      "loss": 0.0009,
      "step": 4807
    },
    {
      "epoch": 6.09,
      "learning_rate": 6.528028933092224e-06,
      "loss": 0.0006,
      "step": 4808
    },
    {
      "epoch": 6.09,
      "learning_rate": 6.518987341772152e-06,
      "loss": 0.0054,
      "step": 4809
    },
    {
      "epoch": 6.09,
      "learning_rate": 6.50994575045208e-06,
      "loss": 0.0004,
      "step": 4810
    },
    {
      "epoch": 6.09,
      "learning_rate": 6.500904159132007e-06,
      "loss": 0.0447,
      "step": 4811
    },
    {
      "epoch": 6.09,
      "learning_rate": 6.491862567811935e-06,
      "loss": 0.0003,
      "step": 4812
    },
    {
      "epoch": 6.09,
      "learning_rate": 6.4828209764918626e-06,
      "loss": 0.0605,
      "step": 4813
    },
    {
      "epoch": 6.09,
      "learning_rate": 6.473779385171791e-06,
      "loss": 0.2232,
      "step": 4814
    },
    {
      "epoch": 6.09,
      "learning_rate": 6.464737793851717e-06,
      "loss": 0.0007,
      "step": 4815
    },
    {
      "epoch": 6.1,
      "learning_rate": 6.455696202531646e-06,
      "loss": 0.0007,
      "step": 4816
    },
    {
      "epoch": 6.1,
      "learning_rate": 6.446654611211574e-06,
      "loss": 0.0004,
      "step": 4817
    },
    {
      "epoch": 6.1,
      "learning_rate": 6.437613019891501e-06,
      "loss": 0.0004,
      "step": 4818
    },
    {
      "epoch": 6.1,
      "learning_rate": 6.428571428571429e-06,
      "loss": 0.0009,
      "step": 4819
    },
    {
      "epoch": 6.1,
      "learning_rate": 6.419529837251356e-06,
      "loss": 0.0022,
      "step": 4820
    },
    {
      "epoch": 6.1,
      "learning_rate": 6.410488245931284e-06,
      "loss": 0.0003,
      "step": 4821
    },
    {
      "epoch": 6.1,
      "learning_rate": 6.4014466546112125e-06,
      "loss": 0.0008,
      "step": 4822
    },
    {
      "epoch": 6.1,
      "learning_rate": 6.392405063291139e-06,
      "loss": 0.0003,
      "step": 4823
    },
    {
      "epoch": 6.11,
      "learning_rate": 6.383363471971067e-06,
      "loss": 0.001,
      "step": 4824
    },
    {
      "epoch": 6.11,
      "learning_rate": 6.374321880650995e-06,
      "loss": 0.0903,
      "step": 4825
    },
    {
      "epoch": 6.11,
      "learning_rate": 6.365280289330923e-06,
      "loss": 0.0004,
      "step": 4826
    },
    {
      "epoch": 6.11,
      "learning_rate": 6.3562386980108494e-06,
      "loss": 0.0003,
      "step": 4827
    },
    {
      "epoch": 6.11,
      "learning_rate": 6.347197106690778e-06,
      "loss": 0.0004,
      "step": 4828
    },
    {
      "epoch": 6.11,
      "learning_rate": 6.338155515370706e-06,
      "loss": 0.0003,
      "step": 4829
    },
    {
      "epoch": 6.11,
      "learning_rate": 6.329113924050633e-06,
      "loss": 0.0005,
      "step": 4830
    },
    {
      "epoch": 6.12,
      "learning_rate": 6.32007233273056e-06,
      "loss": 0.0002,
      "step": 4831
    },
    {
      "epoch": 6.12,
      "learning_rate": 6.311030741410488e-06,
      "loss": 0.0007,
      "step": 4832
    },
    {
      "epoch": 6.12,
      "learning_rate": 6.301989150090416e-06,
      "loss": 0.0006,
      "step": 4833
    },
    {
      "epoch": 6.12,
      "learning_rate": 6.2929475587703446e-06,
      "loss": 0.0002,
      "step": 4834
    },
    {
      "epoch": 6.12,
      "learning_rate": 6.283905967450271e-06,
      "loss": 0.0002,
      "step": 4835
    },
    {
      "epoch": 6.12,
      "learning_rate": 6.274864376130199e-06,
      "loss": 0.0718,
      "step": 4836
    },
    {
      "epoch": 6.12,
      "learning_rate": 6.265822784810127e-06,
      "loss": 0.0017,
      "step": 4837
    },
    {
      "epoch": 6.12,
      "learning_rate": 6.256781193490055e-06,
      "loss": 0.0003,
      "step": 4838
    },
    {
      "epoch": 6.13,
      "learning_rate": 6.247739602169982e-06,
      "loss": 0.0003,
      "step": 4839
    },
    {
      "epoch": 6.13,
      "learning_rate": 6.23869801084991e-06,
      "loss": 0.0004,
      "step": 4840
    },
    {
      "epoch": 6.13,
      "learning_rate": 6.229656419529838e-06,
      "loss": 0.0049,
      "step": 4841
    },
    {
      "epoch": 6.13,
      "learning_rate": 6.220614828209765e-06,
      "loss": 0.0003,
      "step": 4842
    },
    {
      "epoch": 6.13,
      "learning_rate": 6.211573236889693e-06,
      "loss": 0.0002,
      "step": 4843
    },
    {
      "epoch": 6.13,
      "learning_rate": 6.20253164556962e-06,
      "loss": 0.0004,
      "step": 4844
    },
    {
      "epoch": 6.13,
      "learning_rate": 6.193490054249548e-06,
      "loss": 0.0143,
      "step": 4845
    },
    {
      "epoch": 6.13,
      "learning_rate": 6.184448462929476e-06,
      "loss": 0.0003,
      "step": 4846
    },
    {
      "epoch": 6.14,
      "learning_rate": 6.175406871609404e-06,
      "loss": 0.0004,
      "step": 4847
    },
    {
      "epoch": 6.14,
      "learning_rate": 6.1663652802893306e-06,
      "loss": 0.0002,
      "step": 4848
    },
    {
      "epoch": 6.14,
      "learning_rate": 6.157323688969259e-06,
      "loss": 0.0035,
      "step": 4849
    },
    {
      "epoch": 6.14,
      "learning_rate": 6.148282097649186e-06,
      "loss": 0.0004,
      "step": 4850
    },
    {
      "epoch": 6.14,
      "learning_rate": 6.1392405063291144e-06,
      "loss": 0.0658,
      "step": 4851
    },
    {
      "epoch": 6.14,
      "learning_rate": 6.130198915009042e-06,
      "loss": 0.0003,
      "step": 4852
    },
    {
      "epoch": 6.14,
      "learning_rate": 6.12115732368897e-06,
      "loss": 0.0012,
      "step": 4853
    },
    {
      "epoch": 6.14,
      "learning_rate": 6.112115732368897e-06,
      "loss": 0.0014,
      "step": 4854
    },
    {
      "epoch": 6.15,
      "learning_rate": 6.103074141048825e-06,
      "loss": 0.0009,
      "step": 4855
    },
    {
      "epoch": 6.15,
      "learning_rate": 6.094032549728752e-06,
      "loss": 0.0004,
      "step": 4856
    },
    {
      "epoch": 6.15,
      "learning_rate": 6.0849909584086805e-06,
      "loss": 0.0005,
      "step": 4857
    },
    {
      "epoch": 6.15,
      "learning_rate": 6.075949367088608e-06,
      "loss": 0.0004,
      "step": 4858
    },
    {
      "epoch": 6.15,
      "learning_rate": 6.066907775768535e-06,
      "loss": 0.0004,
      "step": 4859
    },
    {
      "epoch": 6.15,
      "learning_rate": 6.057866184448463e-06,
      "loss": 0.0003,
      "step": 4860
    },
    {
      "epoch": 6.15,
      "learning_rate": 6.048824593128391e-06,
      "loss": 0.0006,
      "step": 4861
    },
    {
      "epoch": 6.15,
      "learning_rate": 6.039783001808318e-06,
      "loss": 0.0004,
      "step": 4862
    },
    {
      "epoch": 6.16,
      "learning_rate": 6.0307414104882465e-06,
      "loss": 0.0002,
      "step": 4863
    },
    {
      "epoch": 6.16,
      "learning_rate": 6.021699819168174e-06,
      "loss": 0.0004,
      "step": 4864
    },
    {
      "epoch": 6.16,
      "learning_rate": 6.012658227848101e-06,
      "loss": 0.0495,
      "step": 4865
    },
    {
      "epoch": 6.16,
      "learning_rate": 6.003616636528029e-06,
      "loss": 0.0006,
      "step": 4866
    },
    {
      "epoch": 6.16,
      "learning_rate": 5.994575045207957e-06,
      "loss": 0.0008,
      "step": 4867
    },
    {
      "epoch": 6.16,
      "learning_rate": 5.985533453887884e-06,
      "loss": 0.0004,
      "step": 4868
    },
    {
      "epoch": 6.16,
      "learning_rate": 5.9764918625678126e-06,
      "loss": 0.0003,
      "step": 4869
    },
    {
      "epoch": 6.16,
      "learning_rate": 5.96745027124774e-06,
      "loss": 0.0014,
      "step": 4870
    },
    {
      "epoch": 6.17,
      "learning_rate": 5.958408679927667e-06,
      "loss": 0.0003,
      "step": 4871
    },
    {
      "epoch": 6.17,
      "learning_rate": 5.949367088607595e-06,
      "loss": 0.0005,
      "step": 4872
    },
    {
      "epoch": 6.17,
      "learning_rate": 5.940325497287523e-06,
      "loss": 0.0003,
      "step": 4873
    },
    {
      "epoch": 6.17,
      "learning_rate": 5.93128390596745e-06,
      "loss": 0.0004,
      "step": 4874
    },
    {
      "epoch": 6.17,
      "learning_rate": 5.922242314647379e-06,
      "loss": 0.0003,
      "step": 4875
    },
    {
      "epoch": 6.17,
      "learning_rate": 5.913200723327306e-06,
      "loss": 0.0007,
      "step": 4876
    },
    {
      "epoch": 6.17,
      "learning_rate": 5.904159132007233e-06,
      "loss": 0.0008,
      "step": 4877
    },
    {
      "epoch": 6.17,
      "learning_rate": 5.895117540687161e-06,
      "loss": 0.0253,
      "step": 4878
    },
    {
      "epoch": 6.18,
      "learning_rate": 5.886075949367089e-06,
      "loss": 0.0005,
      "step": 4879
    },
    {
      "epoch": 6.18,
      "learning_rate": 5.877034358047016e-06,
      "loss": 0.0004,
      "step": 4880
    },
    {
      "epoch": 6.18,
      "learning_rate": 5.867992766726945e-06,
      "loss": 0.001,
      "step": 4881
    },
    {
      "epoch": 6.18,
      "learning_rate": 5.858951175406872e-06,
      "loss": 0.0031,
      "step": 4882
    },
    {
      "epoch": 6.18,
      "learning_rate": 5.849909584086799e-06,
      "loss": 0.0318,
      "step": 4883
    },
    {
      "epoch": 6.18,
      "learning_rate": 5.840867992766727e-06,
      "loss": 0.0005,
      "step": 4884
    },
    {
      "epoch": 6.18,
      "learning_rate": 5.831826401446655e-06,
      "loss": 0.0004,
      "step": 4885
    },
    {
      "epoch": 6.18,
      "learning_rate": 5.8227848101265824e-06,
      "loss": 0.0003,
      "step": 4886
    },
    {
      "epoch": 6.19,
      "learning_rate": 5.813743218806511e-06,
      "loss": 0.0003,
      "step": 4887
    },
    {
      "epoch": 6.19,
      "learning_rate": 5.804701627486438e-06,
      "loss": 0.0002,
      "step": 4888
    },
    {
      "epoch": 6.19,
      "learning_rate": 5.7956600361663655e-06,
      "loss": 0.0002,
      "step": 4889
    },
    {
      "epoch": 6.19,
      "learning_rate": 5.786618444846293e-06,
      "loss": 0.0003,
      "step": 4890
    },
    {
      "epoch": 6.19,
      "learning_rate": 5.777576853526221e-06,
      "loss": 0.0004,
      "step": 4891
    },
    {
      "epoch": 6.19,
      "learning_rate": 5.7685352622061485e-06,
      "loss": 0.0003,
      "step": 4892
    },
    {
      "epoch": 6.19,
      "learning_rate": 5.759493670886077e-06,
      "loss": 0.0005,
      "step": 4893
    },
    {
      "epoch": 6.19,
      "learning_rate": 5.750452079566004e-06,
      "loss": 0.0044,
      "step": 4894
    },
    {
      "epoch": 6.2,
      "learning_rate": 5.7414104882459315e-06,
      "loss": 0.0006,
      "step": 4895
    },
    {
      "epoch": 6.2,
      "learning_rate": 5.732368896925859e-06,
      "loss": 0.0007,
      "step": 4896
    },
    {
      "epoch": 6.2,
      "learning_rate": 5.723327305605787e-06,
      "loss": 0.0003,
      "step": 4897
    },
    {
      "epoch": 6.2,
      "learning_rate": 5.7142857142857145e-06,
      "loss": 0.0002,
      "step": 4898
    },
    {
      "epoch": 6.2,
      "learning_rate": 5.705244122965643e-06,
      "loss": 0.0008,
      "step": 4899
    },
    {
      "epoch": 6.2,
      "learning_rate": 5.69620253164557e-06,
      "loss": 0.0009,
      "step": 4900
    },
    {
      "epoch": 6.2,
      "learning_rate": 5.6871609403254975e-06,
      "loss": 0.0006,
      "step": 4901
    },
    {
      "epoch": 6.2,
      "learning_rate": 5.678119349005425e-06,
      "loss": 0.0004,
      "step": 4902
    },
    {
      "epoch": 6.21,
      "learning_rate": 5.669077757685353e-06,
      "loss": 0.0004,
      "step": 4903
    },
    {
      "epoch": 6.21,
      "learning_rate": 5.6600361663652806e-06,
      "loss": 0.0007,
      "step": 4904
    },
    {
      "epoch": 6.21,
      "learning_rate": 5.650994575045209e-06,
      "loss": 0.0008,
      "step": 4905
    },
    {
      "epoch": 6.21,
      "learning_rate": 5.641952983725135e-06,
      "loss": 0.0006,
      "step": 4906
    },
    {
      "epoch": 6.21,
      "learning_rate": 5.6329113924050636e-06,
      "loss": 0.0003,
      "step": 4907
    },
    {
      "epoch": 6.21,
      "learning_rate": 5.623869801084991e-06,
      "loss": 0.0003,
      "step": 4908
    },
    {
      "epoch": 6.21,
      "learning_rate": 5.614828209764919e-06,
      "loss": 0.0004,
      "step": 4909
    },
    {
      "epoch": 6.22,
      "learning_rate": 5.605786618444847e-06,
      "loss": 0.0174,
      "step": 4910
    },
    {
      "epoch": 6.22,
      "learning_rate": 5.596745027124775e-06,
      "loss": 0.0002,
      "step": 4911
    },
    {
      "epoch": 6.22,
      "learning_rate": 5.587703435804701e-06,
      "loss": 0.0008,
      "step": 4912
    },
    {
      "epoch": 6.22,
      "learning_rate": 5.57866184448463e-06,
      "loss": 0.0019,
      "step": 4913
    },
    {
      "epoch": 6.22,
      "learning_rate": 5.569620253164557e-06,
      "loss": 0.0003,
      "step": 4914
    },
    {
      "epoch": 6.22,
      "learning_rate": 5.560578661844485e-06,
      "loss": 0.0009,
      "step": 4915
    },
    {
      "epoch": 6.22,
      "learning_rate": 5.551537070524413e-06,
      "loss": 0.0009,
      "step": 4916
    },
    {
      "epoch": 6.22,
      "learning_rate": 5.54249547920434e-06,
      "loss": 0.0342,
      "step": 4917
    },
    {
      "epoch": 6.23,
      "learning_rate": 5.533453887884267e-06,
      "loss": 0.0017,
      "step": 4918
    },
    {
      "epoch": 6.23,
      "learning_rate": 5.524412296564196e-06,
      "loss": 0.0006,
      "step": 4919
    },
    {
      "epoch": 6.23,
      "learning_rate": 5.515370705244123e-06,
      "loss": 0.0003,
      "step": 4920
    },
    {
      "epoch": 6.23,
      "learning_rate": 5.506329113924051e-06,
      "loss": 0.0003,
      "step": 4921
    },
    {
      "epoch": 6.23,
      "learning_rate": 5.497287522603979e-06,
      "loss": 0.0006,
      "step": 4922
    },
    {
      "epoch": 6.23,
      "learning_rate": 5.488245931283906e-06,
      "loss": 0.0007,
      "step": 4923
    },
    {
      "epoch": 6.23,
      "learning_rate": 5.4792043399638334e-06,
      "loss": 0.0002,
      "step": 4924
    },
    {
      "epoch": 6.23,
      "learning_rate": 5.470162748643762e-06,
      "loss": 0.0006,
      "step": 4925
    },
    {
      "epoch": 6.24,
      "learning_rate": 5.461121157323689e-06,
      "loss": 0.0005,
      "step": 4926
    },
    {
      "epoch": 6.24,
      "learning_rate": 5.452079566003617e-06,
      "loss": 0.0014,
      "step": 4927
    },
    {
      "epoch": 6.24,
      "learning_rate": 5.443037974683545e-06,
      "loss": 0.0007,
      "step": 4928
    },
    {
      "epoch": 6.24,
      "learning_rate": 5.433996383363472e-06,
      "loss": 0.0003,
      "step": 4929
    },
    {
      "epoch": 6.24,
      "learning_rate": 5.4249547920433995e-06,
      "loss": 0.0018,
      "step": 4930
    },
    {
      "epoch": 6.24,
      "learning_rate": 5.415913200723328e-06,
      "loss": 0.0003,
      "step": 4931
    },
    {
      "epoch": 6.24,
      "learning_rate": 5.406871609403255e-06,
      "loss": 0.0004,
      "step": 4932
    },
    {
      "epoch": 6.24,
      "learning_rate": 5.397830018083183e-06,
      "loss": 0.0003,
      "step": 4933
    },
    {
      "epoch": 6.25,
      "learning_rate": 5.388788426763111e-06,
      "loss": 0.0003,
      "step": 4934
    },
    {
      "epoch": 6.25,
      "learning_rate": 5.379746835443038e-06,
      "loss": 0.0002,
      "step": 4935
    },
    {
      "epoch": 6.25,
      "learning_rate": 5.3707052441229655e-06,
      "loss": 0.0005,
      "step": 4936
    },
    {
      "epoch": 6.25,
      "learning_rate": 5.361663652802894e-06,
      "loss": 0.0002,
      "step": 4937
    },
    {
      "epoch": 6.25,
      "learning_rate": 5.352622061482821e-06,
      "loss": 0.0007,
      "step": 4938
    },
    {
      "epoch": 6.25,
      "learning_rate": 5.343580470162749e-06,
      "loss": 0.0005,
      "step": 4939
    },
    {
      "epoch": 6.25,
      "learning_rate": 5.334538878842677e-06,
      "loss": 0.0004,
      "step": 4940
    },
    {
      "epoch": 6.25,
      "learning_rate": 5.325497287522604e-06,
      "loss": 0.1018,
      "step": 4941
    },
    {
      "epoch": 6.26,
      "learning_rate": 5.3164556962025316e-06,
      "loss": 0.0005,
      "step": 4942
    },
    {
      "epoch": 6.26,
      "learning_rate": 5.30741410488246e-06,
      "loss": 0.0003,
      "step": 4943
    },
    {
      "epoch": 6.26,
      "learning_rate": 5.298372513562387e-06,
      "loss": 0.0006,
      "step": 4944
    },
    {
      "epoch": 6.26,
      "learning_rate": 5.2893309222423154e-06,
      "loss": 0.0003,
      "step": 4945
    },
    {
      "epoch": 6.26,
      "learning_rate": 5.280289330922243e-06,
      "loss": 0.0242,
      "step": 4946
    },
    {
      "epoch": 6.26,
      "learning_rate": 5.27124773960217e-06,
      "loss": 0.0005,
      "step": 4947
    },
    {
      "epoch": 6.26,
      "learning_rate": 5.262206148282098e-06,
      "loss": 0.0003,
      "step": 4948
    },
    {
      "epoch": 6.26,
      "learning_rate": 5.253164556962026e-06,
      "loss": 0.0004,
      "step": 4949
    },
    {
      "epoch": 6.27,
      "learning_rate": 5.244122965641953e-06,
      "loss": 0.0005,
      "step": 4950
    },
    {
      "epoch": 6.27,
      "learning_rate": 5.2350813743218815e-06,
      "loss": 0.0003,
      "step": 4951
    },
    {
      "epoch": 6.27,
      "learning_rate": 5.226039783001809e-06,
      "loss": 0.0003,
      "step": 4952
    },
    {
      "epoch": 6.27,
      "learning_rate": 5.216998191681736e-06,
      "loss": 0.0002,
      "step": 4953
    },
    {
      "epoch": 6.27,
      "learning_rate": 5.207956600361664e-06,
      "loss": 0.0003,
      "step": 4954
    },
    {
      "epoch": 6.27,
      "learning_rate": 5.198915009041592e-06,
      "loss": 0.0006,
      "step": 4955
    },
    {
      "epoch": 6.27,
      "learning_rate": 5.189873417721519e-06,
      "loss": 0.0008,
      "step": 4956
    },
    {
      "epoch": 6.27,
      "learning_rate": 5.1808318264014475e-06,
      "loss": 0.0002,
      "step": 4957
    },
    {
      "epoch": 6.28,
      "learning_rate": 5.171790235081374e-06,
      "loss": 0.0004,
      "step": 4958
    },
    {
      "epoch": 6.28,
      "learning_rate": 5.162748643761302e-06,
      "loss": 0.0003,
      "step": 4959
    },
    {
      "epoch": 6.28,
      "learning_rate": 5.15370705244123e-06,
      "loss": 0.0003,
      "step": 4960
    },
    {
      "epoch": 6.28,
      "learning_rate": 5.144665461121158e-06,
      "loss": 0.0003,
      "step": 4961
    },
    {
      "epoch": 6.28,
      "learning_rate": 5.135623869801085e-06,
      "loss": 0.0003,
      "step": 4962
    },
    {
      "epoch": 6.28,
      "learning_rate": 5.1265822784810135e-06,
      "loss": 0.0131,
      "step": 4963
    },
    {
      "epoch": 6.28,
      "learning_rate": 5.11754068716094e-06,
      "loss": 0.0008,
      "step": 4964
    },
    {
      "epoch": 6.28,
      "learning_rate": 5.108499095840868e-06,
      "loss": 0.0003,
      "step": 4965
    },
    {
      "epoch": 6.29,
      "learning_rate": 5.099457504520796e-06,
      "loss": 0.0007,
      "step": 4966
    },
    {
      "epoch": 6.29,
      "learning_rate": 5.090415913200724e-06,
      "loss": 0.0005,
      "step": 4967
    },
    {
      "epoch": 6.29,
      "learning_rate": 5.081374321880651e-06,
      "loss": 0.0005,
      "step": 4968
    },
    {
      "epoch": 6.29,
      "learning_rate": 5.07233273056058e-06,
      "loss": 0.0006,
      "step": 4969
    },
    {
      "epoch": 6.29,
      "learning_rate": 5.063291139240506e-06,
      "loss": 0.0063,
      "step": 4970
    },
    {
      "epoch": 6.29,
      "learning_rate": 5.054249547920434e-06,
      "loss": 0.0007,
      "step": 4971
    },
    {
      "epoch": 6.29,
      "learning_rate": 5.045207956600362e-06,
      "loss": 0.0003,
      "step": 4972
    },
    {
      "epoch": 6.29,
      "learning_rate": 5.03616636528029e-06,
      "loss": 0.0004,
      "step": 4973
    },
    {
      "epoch": 6.3,
      "learning_rate": 5.027124773960217e-06,
      "loss": 0.0004,
      "step": 4974
    },
    {
      "epoch": 6.3,
      "learning_rate": 5.018083182640145e-06,
      "loss": 0.0006,
      "step": 4975
    },
    {
      "epoch": 6.3,
      "learning_rate": 5.009041591320072e-06,
      "loss": 0.0007,
      "step": 4976
    },
    {
      "epoch": 6.3,
      "learning_rate": 5e-06,
      "loss": 0.0005,
      "step": 4977
    },
    {
      "epoch": 6.3,
      "learning_rate": 4.990958408679928e-06,
      "loss": 0.0005,
      "step": 4978
    },
    {
      "epoch": 6.3,
      "learning_rate": 4.981916817359856e-06,
      "loss": 0.0006,
      "step": 4979
    },
    {
      "epoch": 6.3,
      "learning_rate": 4.9728752260397834e-06,
      "loss": 0.001,
      "step": 4980
    },
    {
      "epoch": 6.3,
      "learning_rate": 4.963833634719711e-06,
      "loss": 0.0012,
      "step": 4981
    },
    {
      "epoch": 6.31,
      "learning_rate": 4.954792043399638e-06,
      "loss": 0.0007,
      "step": 4982
    },
    {
      "epoch": 6.31,
      "learning_rate": 4.9457504520795664e-06,
      "loss": 0.0003,
      "step": 4983
    },
    {
      "epoch": 6.31,
      "learning_rate": 4.936708860759494e-06,
      "loss": 0.0003,
      "step": 4984
    },
    {
      "epoch": 6.31,
      "learning_rate": 4.927667269439422e-06,
      "loss": 0.0003,
      "step": 4985
    },
    {
      "epoch": 6.31,
      "learning_rate": 4.9186256781193495e-06,
      "loss": 0.0004,
      "step": 4986
    },
    {
      "epoch": 6.31,
      "learning_rate": 4.909584086799277e-06,
      "loss": 0.0004,
      "step": 4987
    },
    {
      "epoch": 6.31,
      "learning_rate": 4.900542495479204e-06,
      "loss": 0.0057,
      "step": 4988
    },
    {
      "epoch": 6.31,
      "learning_rate": 4.8915009041591325e-06,
      "loss": 0.0008,
      "step": 4989
    },
    {
      "epoch": 6.32,
      "learning_rate": 4.88245931283906e-06,
      "loss": 0.0003,
      "step": 4990
    },
    {
      "epoch": 6.32,
      "learning_rate": 4.873417721518988e-06,
      "loss": 0.012,
      "step": 4991
    },
    {
      "epoch": 6.32,
      "learning_rate": 4.8643761301989155e-06,
      "loss": 0.0004,
      "step": 4992
    },
    {
      "epoch": 6.32,
      "learning_rate": 4.855334538878843e-06,
      "loss": 0.0108,
      "step": 4993
    },
    {
      "epoch": 6.32,
      "learning_rate": 4.84629294755877e-06,
      "loss": 0.0019,
      "step": 4994
    },
    {
      "epoch": 6.32,
      "learning_rate": 4.8372513562386985e-06,
      "loss": 0.0007,
      "step": 4995
    },
    {
      "epoch": 6.32,
      "learning_rate": 4.828209764918626e-06,
      "loss": 0.0005,
      "step": 4996
    },
    {
      "epoch": 6.33,
      "learning_rate": 4.819168173598554e-06,
      "loss": 0.0002,
      "step": 4997
    },
    {
      "epoch": 6.33,
      "learning_rate": 4.8101265822784815e-06,
      "loss": 0.0002,
      "step": 4998
    },
    {
      "epoch": 6.33,
      "learning_rate": 4.801084990958409e-06,
      "loss": 0.0028,
      "step": 4999
    },
    {
      "epoch": 6.33,
      "learning_rate": 4.792043399638336e-06,
      "loss": 0.0002,
      "step": 5000
    },
    {
      "epoch": 6.33,
      "eval_accuracy": 0.9046546546546547,
      "eval_loss": 0.4961221516132355,
      "eval_runtime": 45.1494,
      "eval_samples_per_second": 29.502,
      "eval_steps_per_second": 3.699,
      "step": 5000
    },
    {
      "epoch": 6.33,
      "learning_rate": 4.7830018083182646e-06,
      "loss": 0.0004,
      "step": 5001
    },
    {
      "epoch": 6.33,
      "learning_rate": 4.773960216998192e-06,
      "loss": 0.0002,
      "step": 5002
    },
    {
      "epoch": 6.33,
      "learning_rate": 4.76491862567812e-06,
      "loss": 0.0003,
      "step": 5003
    },
    {
      "epoch": 6.33,
      "learning_rate": 4.755877034358048e-06,
      "loss": 0.0002,
      "step": 5004
    },
    {
      "epoch": 6.34,
      "learning_rate": 4.746835443037975e-06,
      "loss": 0.0004,
      "step": 5005
    },
    {
      "epoch": 6.34,
      "learning_rate": 4.737793851717902e-06,
      "loss": 0.0004,
      "step": 5006
    },
    {
      "epoch": 6.34,
      "learning_rate": 4.728752260397831e-06,
      "loss": 0.0004,
      "step": 5007
    },
    {
      "epoch": 6.34,
      "learning_rate": 4.719710669077758e-06,
      "loss": 0.0005,
      "step": 5008
    },
    {
      "epoch": 6.34,
      "learning_rate": 4.710669077757686e-06,
      "loss": 0.0002,
      "step": 5009
    },
    {
      "epoch": 6.34,
      "learning_rate": 4.701627486437614e-06,
      "loss": 0.0003,
      "step": 5010
    },
    {
      "epoch": 6.34,
      "learning_rate": 4.692585895117541e-06,
      "loss": 0.0004,
      "step": 5011
    },
    {
      "epoch": 6.34,
      "learning_rate": 4.683544303797468e-06,
      "loss": 0.0004,
      "step": 5012
    },
    {
      "epoch": 6.35,
      "learning_rate": 4.674502712477397e-06,
      "loss": 0.0003,
      "step": 5013
    },
    {
      "epoch": 6.35,
      "learning_rate": 4.665461121157324e-06,
      "loss": 0.0002,
      "step": 5014
    },
    {
      "epoch": 6.35,
      "learning_rate": 4.656419529837252e-06,
      "loss": 0.0005,
      "step": 5015
    },
    {
      "epoch": 6.35,
      "learning_rate": 4.647377938517179e-06,
      "loss": 0.0203,
      "step": 5016
    },
    {
      "epoch": 6.35,
      "learning_rate": 4.638336347197107e-06,
      "loss": 0.0003,
      "step": 5017
    },
    {
      "epoch": 6.35,
      "learning_rate": 4.6292947558770344e-06,
      "loss": 0.2105,
      "step": 5018
    },
    {
      "epoch": 6.35,
      "learning_rate": 4.620253164556963e-06,
      "loss": 0.0004,
      "step": 5019
    },
    {
      "epoch": 6.35,
      "learning_rate": 4.61121157323689e-06,
      "loss": 0.0002,
      "step": 5020
    },
    {
      "epoch": 6.36,
      "learning_rate": 4.602169981916818e-06,
      "loss": 0.0002,
      "step": 5021
    },
    {
      "epoch": 6.36,
      "learning_rate": 4.593128390596745e-06,
      "loss": 0.0695,
      "step": 5022
    },
    {
      "epoch": 6.36,
      "learning_rate": 4.584086799276673e-06,
      "loss": 0.0008,
      "step": 5023
    },
    {
      "epoch": 6.36,
      "learning_rate": 4.5750452079566005e-06,
      "loss": 0.0048,
      "step": 5024
    },
    {
      "epoch": 6.36,
      "learning_rate": 4.566003616636529e-06,
      "loss": 0.0006,
      "step": 5025
    },
    {
      "epoch": 6.36,
      "learning_rate": 4.556962025316456e-06,
      "loss": 0.0002,
      "step": 5026
    },
    {
      "epoch": 6.36,
      "learning_rate": 4.547920433996384e-06,
      "loss": 0.0004,
      "step": 5027
    },
    {
      "epoch": 6.36,
      "learning_rate": 4.538878842676311e-06,
      "loss": 0.0005,
      "step": 5028
    },
    {
      "epoch": 6.37,
      "learning_rate": 4.529837251356238e-06,
      "loss": 0.0003,
      "step": 5029
    },
    {
      "epoch": 6.37,
      "learning_rate": 4.5207956600361665e-06,
      "loss": 0.0508,
      "step": 5030
    },
    {
      "epoch": 6.37,
      "learning_rate": 4.511754068716094e-06,
      "loss": 0.0002,
      "step": 5031
    },
    {
      "epoch": 6.37,
      "learning_rate": 4.502712477396022e-06,
      "loss": 0.0003,
      "step": 5032
    },
    {
      "epoch": 6.37,
      "learning_rate": 4.4936708860759495e-06,
      "loss": 0.0776,
      "step": 5033
    },
    {
      "epoch": 6.37,
      "learning_rate": 4.484629294755877e-06,
      "loss": 0.0016,
      "step": 5034
    },
    {
      "epoch": 6.37,
      "learning_rate": 4.475587703435804e-06,
      "loss": 0.0005,
      "step": 5035
    },
    {
      "epoch": 6.37,
      "learning_rate": 4.4665461121157326e-06,
      "loss": 0.0216,
      "step": 5036
    },
    {
      "epoch": 6.38,
      "learning_rate": 4.45750452079566e-06,
      "loss": 0.0319,
      "step": 5037
    },
    {
      "epoch": 6.38,
      "learning_rate": 4.448462929475588e-06,
      "loss": 0.0002,
      "step": 5038
    },
    {
      "epoch": 6.38,
      "learning_rate": 4.439421338155516e-06,
      "loss": 0.0004,
      "step": 5039
    },
    {
      "epoch": 6.38,
      "learning_rate": 4.430379746835443e-06,
      "loss": 0.0017,
      "step": 5040
    },
    {
      "epoch": 6.38,
      "learning_rate": 4.42133815551537e-06,
      "loss": 0.0003,
      "step": 5041
    },
    {
      "epoch": 6.38,
      "learning_rate": 4.412296564195299e-06,
      "loss": 0.0003,
      "step": 5042
    },
    {
      "epoch": 6.38,
      "learning_rate": 4.403254972875226e-06,
      "loss": 0.0024,
      "step": 5043
    },
    {
      "epoch": 6.38,
      "learning_rate": 4.394213381555154e-06,
      "loss": 0.0003,
      "step": 5044
    },
    {
      "epoch": 6.39,
      "learning_rate": 4.385171790235082e-06,
      "loss": 0.0002,
      "step": 5045
    },
    {
      "epoch": 6.39,
      "learning_rate": 4.376130198915009e-06,
      "loss": 0.0003,
      "step": 5046
    },
    {
      "epoch": 6.39,
      "learning_rate": 4.367088607594936e-06,
      "loss": 0.0009,
      "step": 5047
    },
    {
      "epoch": 6.39,
      "learning_rate": 4.358047016274865e-06,
      "loss": 0.0003,
      "step": 5048
    },
    {
      "epoch": 6.39,
      "learning_rate": 4.349005424954792e-06,
      "loss": 0.0003,
      "step": 5049
    },
    {
      "epoch": 6.39,
      "learning_rate": 4.33996383363472e-06,
      "loss": 0.0003,
      "step": 5050
    },
    {
      "epoch": 6.39,
      "learning_rate": 4.330922242314648e-06,
      "loss": 0.1441,
      "step": 5051
    },
    {
      "epoch": 6.39,
      "learning_rate": 4.321880650994575e-06,
      "loss": 0.0002,
      "step": 5052
    },
    {
      "epoch": 6.4,
      "learning_rate": 4.3128390596745024e-06,
      "loss": 0.0004,
      "step": 5053
    },
    {
      "epoch": 6.4,
      "learning_rate": 4.303797468354431e-06,
      "loss": 0.1128,
      "step": 5054
    },
    {
      "epoch": 6.4,
      "learning_rate": 4.294755877034358e-06,
      "loss": 0.0004,
      "step": 5055
    },
    {
      "epoch": 6.4,
      "learning_rate": 4.285714285714286e-06,
      "loss": 0.0003,
      "step": 5056
    },
    {
      "epoch": 6.4,
      "learning_rate": 4.276672694394214e-06,
      "loss": 0.0004,
      "step": 5057
    },
    {
      "epoch": 6.4,
      "learning_rate": 4.267631103074141e-06,
      "loss": 0.0002,
      "step": 5058
    },
    {
      "epoch": 6.4,
      "learning_rate": 4.2585895117540685e-06,
      "loss": 0.0004,
      "step": 5059
    },
    {
      "epoch": 6.4,
      "learning_rate": 4.249547920433997e-06,
      "loss": 0.0002,
      "step": 5060
    },
    {
      "epoch": 6.41,
      "learning_rate": 4.240506329113924e-06,
      "loss": 0.0003,
      "step": 5061
    },
    {
      "epoch": 6.41,
      "learning_rate": 4.231464737793852e-06,
      "loss": 0.0005,
      "step": 5062
    },
    {
      "epoch": 6.41,
      "learning_rate": 4.222423146473779e-06,
      "loss": 0.0003,
      "step": 5063
    },
    {
      "epoch": 6.41,
      "learning_rate": 4.213381555153707e-06,
      "loss": 0.0003,
      "step": 5064
    },
    {
      "epoch": 6.41,
      "learning_rate": 4.2043399638336345e-06,
      "loss": 0.0366,
      "step": 5065
    },
    {
      "epoch": 6.41,
      "learning_rate": 4.195298372513563e-06,
      "loss": 0.0003,
      "step": 5066
    },
    {
      "epoch": 6.41,
      "learning_rate": 4.18625678119349e-06,
      "loss": 0.0179,
      "step": 5067
    },
    {
      "epoch": 6.41,
      "learning_rate": 4.177215189873418e-06,
      "loss": 0.0002,
      "step": 5068
    },
    {
      "epoch": 6.42,
      "learning_rate": 4.168173598553345e-06,
      "loss": 0.0004,
      "step": 5069
    },
    {
      "epoch": 6.42,
      "learning_rate": 4.159132007233273e-06,
      "loss": 0.0008,
      "step": 5070
    },
    {
      "epoch": 6.42,
      "learning_rate": 4.1500904159132006e-06,
      "loss": 0.001,
      "step": 5071
    },
    {
      "epoch": 6.42,
      "learning_rate": 4.141048824593129e-06,
      "loss": 0.0002,
      "step": 5072
    },
    {
      "epoch": 6.42,
      "learning_rate": 4.132007233273056e-06,
      "loss": 0.0003,
      "step": 5073
    },
    {
      "epoch": 6.42,
      "learning_rate": 4.122965641952984e-06,
      "loss": 0.0004,
      "step": 5074
    },
    {
      "epoch": 6.42,
      "learning_rate": 4.113924050632911e-06,
      "loss": 0.0003,
      "step": 5075
    },
    {
      "epoch": 6.43,
      "learning_rate": 4.104882459312839e-06,
      "loss": 0.0014,
      "step": 5076
    },
    {
      "epoch": 6.43,
      "learning_rate": 4.095840867992767e-06,
      "loss": 0.0013,
      "step": 5077
    },
    {
      "epoch": 6.43,
      "learning_rate": 4.086799276672695e-06,
      "loss": 0.0006,
      "step": 5078
    },
    {
      "epoch": 6.43,
      "learning_rate": 4.077757685352622e-06,
      "loss": 0.0206,
      "step": 5079
    },
    {
      "epoch": 6.43,
      "learning_rate": 4.06871609403255e-06,
      "loss": 0.0841,
      "step": 5080
    },
    {
      "epoch": 6.43,
      "learning_rate": 4.059674502712477e-06,
      "loss": 0.0002,
      "step": 5081
    },
    {
      "epoch": 6.43,
      "learning_rate": 4.050632911392405e-06,
      "loss": 0.0003,
      "step": 5082
    },
    {
      "epoch": 6.43,
      "learning_rate": 4.041591320072333e-06,
      "loss": 0.0004,
      "step": 5083
    },
    {
      "epoch": 6.44,
      "learning_rate": 4.032549728752261e-06,
      "loss": 0.0003,
      "step": 5084
    },
    {
      "epoch": 6.44,
      "learning_rate": 4.023508137432188e-06,
      "loss": 0.0003,
      "step": 5085
    },
    {
      "epoch": 6.44,
      "learning_rate": 4.014466546112116e-06,
      "loss": 0.0004,
      "step": 5086
    },
    {
      "epoch": 6.44,
      "learning_rate": 4.005424954792043e-06,
      "loss": 0.0006,
      "step": 5087
    },
    {
      "epoch": 6.44,
      "learning_rate": 3.996383363471971e-06,
      "loss": 0.0003,
      "step": 5088
    },
    {
      "epoch": 6.44,
      "learning_rate": 3.987341772151899e-06,
      "loss": 0.0005,
      "step": 5089
    },
    {
      "epoch": 6.44,
      "learning_rate": 3.978300180831827e-06,
      "loss": 0.0004,
      "step": 5090
    },
    {
      "epoch": 6.44,
      "learning_rate": 3.969258589511754e-06,
      "loss": 0.0006,
      "step": 5091
    },
    {
      "epoch": 6.45,
      "learning_rate": 3.960216998191682e-06,
      "loss": 0.0358,
      "step": 5092
    },
    {
      "epoch": 6.45,
      "learning_rate": 3.951175406871609e-06,
      "loss": 0.2203,
      "step": 5093
    },
    {
      "epoch": 6.45,
      "learning_rate": 3.942133815551537e-06,
      "loss": 0.0007,
      "step": 5094
    },
    {
      "epoch": 6.45,
      "learning_rate": 3.933092224231465e-06,
      "loss": 0.0002,
      "step": 5095
    },
    {
      "epoch": 6.45,
      "learning_rate": 3.924050632911393e-06,
      "loss": 0.0005,
      "step": 5096
    },
    {
      "epoch": 6.45,
      "learning_rate": 3.91500904159132e-06,
      "loss": 0.001,
      "step": 5097
    },
    {
      "epoch": 6.45,
      "learning_rate": 3.905967450271248e-06,
      "loss": 0.0003,
      "step": 5098
    },
    {
      "epoch": 6.45,
      "learning_rate": 3.896925858951175e-06,
      "loss": 0.0003,
      "step": 5099
    },
    {
      "epoch": 6.46,
      "learning_rate": 3.887884267631103e-06,
      "loss": 0.0003,
      "step": 5100
    },
    {
      "epoch": 6.46,
      "learning_rate": 3.878842676311031e-06,
      "loss": 0.0004,
      "step": 5101
    },
    {
      "epoch": 6.46,
      "learning_rate": 3.869801084990959e-06,
      "loss": 0.0004,
      "step": 5102
    },
    {
      "epoch": 6.46,
      "learning_rate": 3.860759493670886e-06,
      "loss": 0.0003,
      "step": 5103
    },
    {
      "epoch": 6.46,
      "learning_rate": 3.851717902350814e-06,
      "loss": 0.0004,
      "step": 5104
    },
    {
      "epoch": 6.46,
      "learning_rate": 3.842676311030741e-06,
      "loss": 0.0003,
      "step": 5105
    },
    {
      "epoch": 6.46,
      "learning_rate": 3.833634719710669e-06,
      "loss": 0.0003,
      "step": 5106
    },
    {
      "epoch": 6.46,
      "learning_rate": 3.824593128390597e-06,
      "loss": 0.0003,
      "step": 5107
    },
    {
      "epoch": 6.47,
      "learning_rate": 3.815551537070525e-06,
      "loss": 0.0004,
      "step": 5108
    },
    {
      "epoch": 6.47,
      "learning_rate": 3.806509945750452e-06,
      "loss": 0.0004,
      "step": 5109
    },
    {
      "epoch": 6.47,
      "learning_rate": 3.7974683544303802e-06,
      "loss": 0.0003,
      "step": 5110
    },
    {
      "epoch": 6.47,
      "learning_rate": 3.788426763110307e-06,
      "loss": 0.0008,
      "step": 5111
    },
    {
      "epoch": 6.47,
      "learning_rate": 3.7793851717902354e-06,
      "loss": 0.0006,
      "step": 5112
    },
    {
      "epoch": 6.47,
      "learning_rate": 3.770343580470163e-06,
      "loss": 0.0482,
      "step": 5113
    },
    {
      "epoch": 6.47,
      "learning_rate": 3.7613019891500906e-06,
      "loss": 0.0005,
      "step": 5114
    },
    {
      "epoch": 6.47,
      "learning_rate": 3.752260397830018e-06,
      "loss": 0.013,
      "step": 5115
    },
    {
      "epoch": 6.48,
      "learning_rate": 3.743218806509946e-06,
      "loss": 0.0002,
      "step": 5116
    },
    {
      "epoch": 6.48,
      "learning_rate": 3.7341772151898732e-06,
      "loss": 0.001,
      "step": 5117
    },
    {
      "epoch": 6.48,
      "learning_rate": 3.7251356238698015e-06,
      "loss": 0.0003,
      "step": 5118
    },
    {
      "epoch": 6.48,
      "learning_rate": 3.716094032549729e-06,
      "loss": 0.0077,
      "step": 5119
    },
    {
      "epoch": 6.48,
      "learning_rate": 3.7070524412296567e-06,
      "loss": 0.0003,
      "step": 5120
    },
    {
      "epoch": 6.48,
      "learning_rate": 3.698010849909584e-06,
      "loss": 0.0011,
      "step": 5121
    },
    {
      "epoch": 6.48,
      "learning_rate": 3.688969258589512e-06,
      "loss": 0.0003,
      "step": 5122
    },
    {
      "epoch": 6.48,
      "learning_rate": 3.6799276672694393e-06,
      "loss": 0.1437,
      "step": 5123
    },
    {
      "epoch": 6.49,
      "learning_rate": 3.6708860759493675e-06,
      "loss": 0.0005,
      "step": 5124
    },
    {
      "epoch": 6.49,
      "learning_rate": 3.661844484629295e-06,
      "loss": 0.0004,
      "step": 5125
    },
    {
      "epoch": 6.49,
      "learning_rate": 3.6528028933092227e-06,
      "loss": 0.0007,
      "step": 5126
    },
    {
      "epoch": 6.49,
      "learning_rate": 3.64376130198915e-06,
      "loss": 0.0005,
      "step": 5127
    },
    {
      "epoch": 6.49,
      "learning_rate": 3.634719710669078e-06,
      "loss": 0.0009,
      "step": 5128
    },
    {
      "epoch": 6.49,
      "learning_rate": 3.6256781193490053e-06,
      "loss": 0.0004,
      "step": 5129
    },
    {
      "epoch": 6.49,
      "learning_rate": 3.6166365280289336e-06,
      "loss": 0.0004,
      "step": 5130
    },
    {
      "epoch": 6.49,
      "learning_rate": 3.6075949367088605e-06,
      "loss": 0.0003,
      "step": 5131
    },
    {
      "epoch": 6.5,
      "learning_rate": 3.5985533453887888e-06,
      "loss": 0.0013,
      "step": 5132
    },
    {
      "epoch": 6.5,
      "learning_rate": 3.589511754068716e-06,
      "loss": 0.0008,
      "step": 5133
    },
    {
      "epoch": 6.5,
      "learning_rate": 3.580470162748644e-06,
      "loss": 0.0011,
      "step": 5134
    },
    {
      "epoch": 6.5,
      "learning_rate": 3.5714285714285714e-06,
      "loss": 0.0006,
      "step": 5135
    },
    {
      "epoch": 6.5,
      "learning_rate": 3.5623869801084996e-06,
      "loss": 0.0003,
      "step": 5136
    },
    {
      "epoch": 6.5,
      "learning_rate": 3.5533453887884266e-06,
      "loss": 0.0004,
      "step": 5137
    },
    {
      "epoch": 6.5,
      "learning_rate": 3.544303797468355e-06,
      "loss": 0.0008,
      "step": 5138
    },
    {
      "epoch": 6.5,
      "learning_rate": 3.535262206148282e-06,
      "loss": 0.0003,
      "step": 5139
    },
    {
      "epoch": 6.51,
      "learning_rate": 3.52622061482821e-06,
      "loss": 0.0002,
      "step": 5140
    },
    {
      "epoch": 6.51,
      "learning_rate": 3.5171790235081374e-06,
      "loss": 0.0004,
      "step": 5141
    },
    {
      "epoch": 6.51,
      "learning_rate": 3.5081374321880656e-06,
      "loss": 0.0004,
      "step": 5142
    },
    {
      "epoch": 6.51,
      "learning_rate": 3.4990958408679926e-06,
      "loss": 0.0075,
      "step": 5143
    },
    {
      "epoch": 6.51,
      "learning_rate": 3.490054249547921e-06,
      "loss": 0.0005,
      "step": 5144
    },
    {
      "epoch": 6.51,
      "learning_rate": 3.4810126582278482e-06,
      "loss": 0.0005,
      "step": 5145
    },
    {
      "epoch": 6.51,
      "learning_rate": 3.471971066907776e-06,
      "loss": 0.0002,
      "step": 5146
    },
    {
      "epoch": 6.51,
      "learning_rate": 3.4629294755877034e-06,
      "loss": 0.0002,
      "step": 5147
    },
    {
      "epoch": 6.52,
      "learning_rate": 3.4538878842676313e-06,
      "loss": 0.0003,
      "step": 5148
    },
    {
      "epoch": 6.52,
      "learning_rate": 3.4448462929475586e-06,
      "loss": 0.001,
      "step": 5149
    },
    {
      "epoch": 6.52,
      "learning_rate": 3.435804701627487e-06,
      "loss": 0.0006,
      "step": 5150
    },
    {
      "epoch": 6.52,
      "learning_rate": 3.4267631103074143e-06,
      "loss": 0.0003,
      "step": 5151
    },
    {
      "epoch": 6.52,
      "learning_rate": 3.417721518987342e-06,
      "loss": 0.0003,
      "step": 5152
    },
    {
      "epoch": 6.52,
      "learning_rate": 3.4086799276672695e-06,
      "loss": 0.0052,
      "step": 5153
    },
    {
      "epoch": 6.52,
      "learning_rate": 3.3996383363471973e-06,
      "loss": 0.0003,
      "step": 5154
    },
    {
      "epoch": 6.52,
      "learning_rate": 3.3905967450271247e-06,
      "loss": 0.0005,
      "step": 5155
    },
    {
      "epoch": 6.53,
      "learning_rate": 3.381555153707053e-06,
      "loss": 0.0005,
      "step": 5156
    },
    {
      "epoch": 6.53,
      "learning_rate": 3.3725135623869803e-06,
      "loss": 0.0003,
      "step": 5157
    },
    {
      "epoch": 6.53,
      "learning_rate": 3.363471971066908e-06,
      "loss": 0.0017,
      "step": 5158
    },
    {
      "epoch": 6.53,
      "learning_rate": 3.3544303797468355e-06,
      "loss": 0.0009,
      "step": 5159
    },
    {
      "epoch": 6.53,
      "learning_rate": 3.3453887884267633e-06,
      "loss": 0.0684,
      "step": 5160
    },
    {
      "epoch": 6.53,
      "learning_rate": 3.3363471971066907e-06,
      "loss": 0.041,
      "step": 5161
    },
    {
      "epoch": 6.53,
      "learning_rate": 3.327305605786619e-06,
      "loss": 0.0003,
      "step": 5162
    },
    {
      "epoch": 6.54,
      "learning_rate": 3.318264014466546e-06,
      "loss": 0.0005,
      "step": 5163
    },
    {
      "epoch": 6.54,
      "learning_rate": 3.309222423146474e-06,
      "loss": 0.0003,
      "step": 5164
    },
    {
      "epoch": 6.54,
      "learning_rate": 3.3001808318264016e-06,
      "loss": 0.0003,
      "step": 5165
    },
    {
      "epoch": 6.54,
      "learning_rate": 3.2911392405063294e-06,
      "loss": 0.0004,
      "step": 5166
    },
    {
      "epoch": 6.54,
      "learning_rate": 3.2820976491862568e-06,
      "loss": 0.0002,
      "step": 5167
    },
    {
      "epoch": 6.54,
      "learning_rate": 3.273056057866185e-06,
      "loss": 0.0156,
      "step": 5168
    },
    {
      "epoch": 6.54,
      "learning_rate": 3.264014466546112e-06,
      "loss": 0.0005,
      "step": 5169
    },
    {
      "epoch": 6.54,
      "learning_rate": 3.25497287522604e-06,
      "loss": 0.0027,
      "step": 5170
    },
    {
      "epoch": 6.55,
      "learning_rate": 3.2459312839059676e-06,
      "loss": 0.0006,
      "step": 5171
    },
    {
      "epoch": 6.55,
      "learning_rate": 3.2368896925858954e-06,
      "loss": 0.0004,
      "step": 5172
    },
    {
      "epoch": 6.55,
      "learning_rate": 3.227848101265823e-06,
      "loss": 0.0037,
      "step": 5173
    },
    {
      "epoch": 6.55,
      "learning_rate": 3.2188065099457506e-06,
      "loss": 0.0004,
      "step": 5174
    },
    {
      "epoch": 6.55,
      "learning_rate": 3.209764918625678e-06,
      "loss": 0.0004,
      "step": 5175
    },
    {
      "epoch": 6.55,
      "learning_rate": 3.2007233273056062e-06,
      "loss": 0.0002,
      "step": 5176
    },
    {
      "epoch": 6.55,
      "learning_rate": 3.1916817359855336e-06,
      "loss": 0.0005,
      "step": 5177
    },
    {
      "epoch": 6.55,
      "learning_rate": 3.1826401446654614e-06,
      "loss": 0.0005,
      "step": 5178
    },
    {
      "epoch": 6.56,
      "learning_rate": 3.173598553345389e-06,
      "loss": 0.0005,
      "step": 5179
    },
    {
      "epoch": 6.56,
      "learning_rate": 3.1645569620253167e-06,
      "loss": 0.0004,
      "step": 5180
    },
    {
      "epoch": 6.56,
      "learning_rate": 3.155515370705244e-06,
      "loss": 0.0005,
      "step": 5181
    },
    {
      "epoch": 6.56,
      "learning_rate": 3.1464737793851723e-06,
      "loss": 0.0002,
      "step": 5182
    },
    {
      "epoch": 6.56,
      "learning_rate": 3.1374321880650997e-06,
      "loss": 0.0005,
      "step": 5183
    },
    {
      "epoch": 6.56,
      "learning_rate": 3.1283905967450275e-06,
      "loss": 0.0002,
      "step": 5184
    },
    {
      "epoch": 6.56,
      "learning_rate": 3.119349005424955e-06,
      "loss": 0.0009,
      "step": 5185
    },
    {
      "epoch": 6.56,
      "learning_rate": 3.1103074141048827e-06,
      "loss": 0.0854,
      "step": 5186
    },
    {
      "epoch": 6.57,
      "learning_rate": 3.10126582278481e-06,
      "loss": 0.0428,
      "step": 5187
    },
    {
      "epoch": 6.57,
      "learning_rate": 3.092224231464738e-06,
      "loss": 0.0003,
      "step": 5188
    },
    {
      "epoch": 6.57,
      "learning_rate": 3.0831826401446653e-06,
      "loss": 0.0004,
      "step": 5189
    },
    {
      "epoch": 6.57,
      "learning_rate": 3.074141048824593e-06,
      "loss": 0.0012,
      "step": 5190
    },
    {
      "epoch": 6.57,
      "learning_rate": 3.065099457504521e-06,
      "loss": 0.0003,
      "step": 5191
    },
    {
      "epoch": 6.57,
      "learning_rate": 3.0560578661844483e-06,
      "loss": 0.001,
      "step": 5192
    },
    {
      "epoch": 6.57,
      "learning_rate": 3.047016274864376e-06,
      "loss": 0.1354,
      "step": 5193
    },
    {
      "epoch": 6.57,
      "learning_rate": 3.037974683544304e-06,
      "loss": 0.0005,
      "step": 5194
    },
    {
      "epoch": 6.58,
      "learning_rate": 3.0289330922242313e-06,
      "loss": 0.0006,
      "step": 5195
    },
    {
      "epoch": 6.58,
      "learning_rate": 3.019891500904159e-06,
      "loss": 0.0003,
      "step": 5196
    },
    {
      "epoch": 6.58,
      "learning_rate": 3.010849909584087e-06,
      "loss": 0.0004,
      "step": 5197
    },
    {
      "epoch": 6.58,
      "learning_rate": 3.0018083182640143e-06,
      "loss": 0.0003,
      "step": 5198
    },
    {
      "epoch": 6.58,
      "learning_rate": 2.992766726943942e-06,
      "loss": 0.0004,
      "step": 5199
    },
    {
      "epoch": 6.58,
      "learning_rate": 2.98372513562387e-06,
      "loss": 0.0014,
      "step": 5200
    },
    {
      "epoch": 6.58,
      "learning_rate": 2.9746835443037974e-06,
      "loss": 0.0041,
      "step": 5201
    },
    {
      "epoch": 6.58,
      "learning_rate": 2.965641952983725e-06,
      "loss": 0.0004,
      "step": 5202
    },
    {
      "epoch": 6.59,
      "learning_rate": 2.956600361663653e-06,
      "loss": 0.0006,
      "step": 5203
    },
    {
      "epoch": 6.59,
      "learning_rate": 2.9475587703435804e-06,
      "loss": 0.0011,
      "step": 5204
    },
    {
      "epoch": 6.59,
      "learning_rate": 2.938517179023508e-06,
      "loss": 0.0003,
      "step": 5205
    },
    {
      "epoch": 6.59,
      "learning_rate": 2.929475587703436e-06,
      "loss": 0.0003,
      "step": 5206
    },
    {
      "epoch": 6.59,
      "learning_rate": 2.9204339963833634e-06,
      "loss": 0.0189,
      "step": 5207
    },
    {
      "epoch": 6.59,
      "learning_rate": 2.9113924050632912e-06,
      "loss": 0.0005,
      "step": 5208
    },
    {
      "epoch": 6.59,
      "learning_rate": 2.902350813743219e-06,
      "loss": 0.0003,
      "step": 5209
    },
    {
      "epoch": 6.59,
      "learning_rate": 2.8933092224231464e-06,
      "loss": 0.0002,
      "step": 5210
    },
    {
      "epoch": 6.6,
      "learning_rate": 2.8842676311030742e-06,
      "loss": 0.0006,
      "step": 5211
    },
    {
      "epoch": 6.6,
      "learning_rate": 2.875226039783002e-06,
      "loss": 0.0005,
      "step": 5212
    },
    {
      "epoch": 6.6,
      "learning_rate": 2.8661844484629294e-06,
      "loss": 0.0014,
      "step": 5213
    },
    {
      "epoch": 6.6,
      "learning_rate": 2.8571428571428573e-06,
      "loss": 0.0474,
      "step": 5214
    },
    {
      "epoch": 6.6,
      "learning_rate": 2.848101265822785e-06,
      "loss": 0.0003,
      "step": 5215
    },
    {
      "epoch": 6.6,
      "learning_rate": 2.8390596745027125e-06,
      "loss": 0.0018,
      "step": 5216
    },
    {
      "epoch": 6.6,
      "learning_rate": 2.8300180831826403e-06,
      "loss": 0.0005,
      "step": 5217
    },
    {
      "epoch": 6.6,
      "learning_rate": 2.8209764918625677e-06,
      "loss": 0.0007,
      "step": 5218
    },
    {
      "epoch": 6.61,
      "learning_rate": 2.8119349005424955e-06,
      "loss": 0.0003,
      "step": 5219
    },
    {
      "epoch": 6.61,
      "learning_rate": 2.8028933092224233e-06,
      "loss": 0.0556,
      "step": 5220
    },
    {
      "epoch": 6.61,
      "learning_rate": 2.7938517179023507e-06,
      "loss": 0.0004,
      "step": 5221
    },
    {
      "epoch": 6.61,
      "learning_rate": 2.7848101265822785e-06,
      "loss": 0.0003,
      "step": 5222
    },
    {
      "epoch": 6.61,
      "learning_rate": 2.7757685352622063e-06,
      "loss": 0.0082,
      "step": 5223
    },
    {
      "epoch": 6.61,
      "learning_rate": 2.7667269439421337e-06,
      "loss": 0.0007,
      "step": 5224
    },
    {
      "epoch": 6.61,
      "learning_rate": 2.7576853526220615e-06,
      "loss": 0.0003,
      "step": 5225
    },
    {
      "epoch": 6.61,
      "learning_rate": 2.7486437613019893e-06,
      "loss": 0.2699,
      "step": 5226
    },
    {
      "epoch": 6.62,
      "learning_rate": 2.7396021699819167e-06,
      "loss": 0.0639,
      "step": 5227
    },
    {
      "epoch": 6.62,
      "learning_rate": 2.7305605786618445e-06,
      "loss": 0.001,
      "step": 5228
    },
    {
      "epoch": 6.62,
      "learning_rate": 2.7215189873417724e-06,
      "loss": 0.0003,
      "step": 5229
    },
    {
      "epoch": 6.62,
      "learning_rate": 2.7124773960216997e-06,
      "loss": 0.0004,
      "step": 5230
    },
    {
      "epoch": 6.62,
      "learning_rate": 2.7034358047016276e-06,
      "loss": 0.0004,
      "step": 5231
    },
    {
      "epoch": 6.62,
      "learning_rate": 2.6943942133815554e-06,
      "loss": 0.0005,
      "step": 5232
    },
    {
      "epoch": 6.62,
      "learning_rate": 2.6853526220614828e-06,
      "loss": 0.0143,
      "step": 5233
    },
    {
      "epoch": 6.62,
      "learning_rate": 2.6763110307414106e-06,
      "loss": 0.0006,
      "step": 5234
    },
    {
      "epoch": 6.63,
      "learning_rate": 2.6672694394213384e-06,
      "loss": 0.001,
      "step": 5235
    },
    {
      "epoch": 6.63,
      "learning_rate": 2.6582278481012658e-06,
      "loss": 0.0003,
      "step": 5236
    },
    {
      "epoch": 6.63,
      "learning_rate": 2.6491862567811936e-06,
      "loss": 0.0003,
      "step": 5237
    },
    {
      "epoch": 6.63,
      "learning_rate": 2.6401446654611214e-06,
      "loss": 0.0009,
      "step": 5238
    },
    {
      "epoch": 6.63,
      "learning_rate": 2.631103074141049e-06,
      "loss": 0.0003,
      "step": 5239
    },
    {
      "epoch": 6.63,
      "learning_rate": 2.6220614828209766e-06,
      "loss": 0.0013,
      "step": 5240
    },
    {
      "epoch": 6.63,
      "learning_rate": 2.6130198915009044e-06,
      "loss": 0.0004,
      "step": 5241
    },
    {
      "epoch": 6.64,
      "learning_rate": 2.603978300180832e-06,
      "loss": 0.0005,
      "step": 5242
    },
    {
      "epoch": 6.64,
      "learning_rate": 2.5949367088607596e-06,
      "loss": 0.0003,
      "step": 5243
    },
    {
      "epoch": 6.64,
      "learning_rate": 2.585895117540687e-06,
      "loss": 0.0003,
      "step": 5244
    },
    {
      "epoch": 6.64,
      "learning_rate": 2.576853526220615e-06,
      "loss": 0.0004,
      "step": 5245
    },
    {
      "epoch": 6.64,
      "learning_rate": 2.5678119349005427e-06,
      "loss": 0.0016,
      "step": 5246
    },
    {
      "epoch": 6.64,
      "learning_rate": 2.55877034358047e-06,
      "loss": 0.0005,
      "step": 5247
    },
    {
      "epoch": 6.64,
      "learning_rate": 2.549728752260398e-06,
      "loss": 0.0003,
      "step": 5248
    },
    {
      "epoch": 6.64,
      "learning_rate": 2.5406871609403257e-06,
      "loss": 0.0002,
      "step": 5249
    },
    {
      "epoch": 6.65,
      "learning_rate": 2.531645569620253e-06,
      "loss": 0.0004,
      "step": 5250
    },
    {
      "epoch": 6.65,
      "eval_accuracy": 0.9069069069069069,
      "eval_loss": 0.49461740255355835,
      "eval_runtime": 45.1019,
      "eval_samples_per_second": 29.533,
      "eval_steps_per_second": 3.703,
      "step": 5250
    },
    {
      "epoch": 6.65,
      "learning_rate": 2.522603978300181e-06,
      "loss": 0.0003,
      "step": 5251
    },
    {
      "epoch": 6.65,
      "learning_rate": 2.5135623869801087e-06,
      "loss": 0.1441,
      "step": 5252
    },
    {
      "epoch": 6.65,
      "learning_rate": 2.504520795660036e-06,
      "loss": 0.0004,
      "step": 5253
    },
    {
      "epoch": 6.65,
      "learning_rate": 2.495479204339964e-06,
      "loss": 0.0002,
      "step": 5254
    },
    {
      "epoch": 6.65,
      "learning_rate": 2.4864376130198917e-06,
      "loss": 0.0114,
      "step": 5255
    },
    {
      "epoch": 6.65,
      "learning_rate": 2.477396021699819e-06,
      "loss": 0.0002,
      "step": 5256
    },
    {
      "epoch": 6.65,
      "learning_rate": 2.468354430379747e-06,
      "loss": 0.0004,
      "step": 5257
    },
    {
      "epoch": 6.66,
      "learning_rate": 2.4593128390596747e-06,
      "loss": 0.0004,
      "step": 5258
    },
    {
      "epoch": 6.66,
      "learning_rate": 2.450271247739602e-06,
      "loss": 0.0023,
      "step": 5259
    },
    {
      "epoch": 6.66,
      "learning_rate": 2.44122965641953e-06,
      "loss": 0.0005,
      "step": 5260
    },
    {
      "epoch": 6.66,
      "learning_rate": 2.4321880650994578e-06,
      "loss": 0.0004,
      "step": 5261
    },
    {
      "epoch": 6.66,
      "learning_rate": 2.423146473779385e-06,
      "loss": 0.0003,
      "step": 5262
    },
    {
      "epoch": 6.66,
      "learning_rate": 2.414104882459313e-06,
      "loss": 0.0004,
      "step": 5263
    },
    {
      "epoch": 6.66,
      "learning_rate": 2.4050632911392408e-06,
      "loss": 0.0004,
      "step": 5264
    },
    {
      "epoch": 6.66,
      "learning_rate": 2.396021699819168e-06,
      "loss": 0.0003,
      "step": 5265
    },
    {
      "epoch": 6.67,
      "learning_rate": 2.386980108499096e-06,
      "loss": 0.0003,
      "step": 5266
    },
    {
      "epoch": 6.67,
      "learning_rate": 2.377938517179024e-06,
      "loss": 0.0002,
      "step": 5267
    },
    {
      "epoch": 6.67,
      "learning_rate": 2.368896925858951e-06,
      "loss": 0.0004,
      "step": 5268
    },
    {
      "epoch": 6.67,
      "learning_rate": 2.359855334538879e-06,
      "loss": 0.0007,
      "step": 5269
    },
    {
      "epoch": 6.67,
      "learning_rate": 2.350813743218807e-06,
      "loss": 0.0183,
      "step": 5270
    },
    {
      "epoch": 6.67,
      "learning_rate": 2.341772151898734e-06,
      "loss": 0.0002,
      "step": 5271
    },
    {
      "epoch": 6.67,
      "learning_rate": 2.332730560578662e-06,
      "loss": 0.0003,
      "step": 5272
    },
    {
      "epoch": 6.67,
      "learning_rate": 2.3236889692585894e-06,
      "loss": 0.001,
      "step": 5273
    },
    {
      "epoch": 6.68,
      "learning_rate": 2.3146473779385172e-06,
      "loss": 0.0003,
      "step": 5274
    },
    {
      "epoch": 6.68,
      "learning_rate": 2.305605786618445e-06,
      "loss": 0.1055,
      "step": 5275
    },
    {
      "epoch": 6.68,
      "learning_rate": 2.2965641952983724e-06,
      "loss": 0.001,
      "step": 5276
    },
    {
      "epoch": 6.68,
      "learning_rate": 2.2875226039783002e-06,
      "loss": 0.0006,
      "step": 5277
    },
    {
      "epoch": 6.68,
      "learning_rate": 2.278481012658228e-06,
      "loss": 0.0007,
      "step": 5278
    },
    {
      "epoch": 6.68,
      "learning_rate": 2.2694394213381554e-06,
      "loss": 0.0031,
      "step": 5279
    },
    {
      "epoch": 6.68,
      "learning_rate": 2.2603978300180833e-06,
      "loss": 0.0004,
      "step": 5280
    },
    {
      "epoch": 6.68,
      "learning_rate": 2.251356238698011e-06,
      "loss": 0.0013,
      "step": 5281
    },
    {
      "epoch": 6.69,
      "learning_rate": 2.2423146473779385e-06,
      "loss": 0.0009,
      "step": 5282
    },
    {
      "epoch": 6.69,
      "learning_rate": 2.2332730560578663e-06,
      "loss": 0.0012,
      "step": 5283
    },
    {
      "epoch": 6.69,
      "learning_rate": 2.224231464737794e-06,
      "loss": 0.0005,
      "step": 5284
    },
    {
      "epoch": 6.69,
      "learning_rate": 2.2151898734177215e-06,
      "loss": 0.0012,
      "step": 5285
    },
    {
      "epoch": 6.69,
      "learning_rate": 2.2061482820976493e-06,
      "loss": 0.0002,
      "step": 5286
    },
    {
      "epoch": 6.69,
      "learning_rate": 2.197106690777577e-06,
      "loss": 0.0002,
      "step": 5287
    },
    {
      "epoch": 6.69,
      "learning_rate": 2.1880650994575045e-06,
      "loss": 0.0002,
      "step": 5288
    },
    {
      "epoch": 6.69,
      "learning_rate": 2.1790235081374323e-06,
      "loss": 0.0003,
      "step": 5289
    },
    {
      "epoch": 6.7,
      "learning_rate": 2.16998191681736e-06,
      "loss": 0.0002,
      "step": 5290
    },
    {
      "epoch": 6.7,
      "learning_rate": 2.1609403254972875e-06,
      "loss": 0.0002,
      "step": 5291
    },
    {
      "epoch": 6.7,
      "learning_rate": 2.1518987341772153e-06,
      "loss": 0.0003,
      "step": 5292
    },
    {
      "epoch": 6.7,
      "learning_rate": 2.142857142857143e-06,
      "loss": 0.0005,
      "step": 5293
    },
    {
      "epoch": 6.7,
      "learning_rate": 2.1338155515370705e-06,
      "loss": 0.0005,
      "step": 5294
    },
    {
      "epoch": 6.7,
      "learning_rate": 2.1247739602169984e-06,
      "loss": 0.0003,
      "step": 5295
    },
    {
      "epoch": 6.7,
      "learning_rate": 2.115732368896926e-06,
      "loss": 0.0004,
      "step": 5296
    },
    {
      "epoch": 6.7,
      "learning_rate": 2.1066907775768536e-06,
      "loss": 0.0645,
      "step": 5297
    },
    {
      "epoch": 6.71,
      "learning_rate": 2.0976491862567814e-06,
      "loss": 0.0004,
      "step": 5298
    },
    {
      "epoch": 6.71,
      "learning_rate": 2.088607594936709e-06,
      "loss": 0.0008,
      "step": 5299
    },
    {
      "epoch": 6.71,
      "learning_rate": 2.0795660036166366e-06,
      "loss": 0.0002,
      "step": 5300
    },
    {
      "epoch": 6.71,
      "learning_rate": 2.0705244122965644e-06,
      "loss": 0.0002,
      "step": 5301
    },
    {
      "epoch": 6.71,
      "learning_rate": 2.061482820976492e-06,
      "loss": 0.001,
      "step": 5302
    },
    {
      "epoch": 6.71,
      "learning_rate": 2.0524412296564196e-06,
      "loss": 0.0002,
      "step": 5303
    },
    {
      "epoch": 6.71,
      "learning_rate": 2.0433996383363474e-06,
      "loss": 0.002,
      "step": 5304
    },
    {
      "epoch": 6.71,
      "learning_rate": 2.034358047016275e-06,
      "loss": 0.0003,
      "step": 5305
    },
    {
      "epoch": 6.72,
      "learning_rate": 2.0253164556962026e-06,
      "loss": 0.0003,
      "step": 5306
    },
    {
      "epoch": 6.72,
      "learning_rate": 2.0162748643761304e-06,
      "loss": 0.0004,
      "step": 5307
    },
    {
      "epoch": 6.72,
      "learning_rate": 2.007233273056058e-06,
      "loss": 0.0007,
      "step": 5308
    },
    {
      "epoch": 6.72,
      "learning_rate": 1.9981916817359856e-06,
      "loss": 0.0003,
      "step": 5309
    },
    {
      "epoch": 6.72,
      "learning_rate": 1.9891500904159135e-06,
      "loss": 0.0003,
      "step": 5310
    },
    {
      "epoch": 6.72,
      "learning_rate": 1.980108499095841e-06,
      "loss": 0.0008,
      "step": 5311
    },
    {
      "epoch": 6.72,
      "learning_rate": 1.9710669077757687e-06,
      "loss": 0.0003,
      "step": 5312
    },
    {
      "epoch": 6.72,
      "learning_rate": 1.9620253164556965e-06,
      "loss": 0.0005,
      "step": 5313
    },
    {
      "epoch": 6.73,
      "learning_rate": 1.952983725135624e-06,
      "loss": 0.0004,
      "step": 5314
    },
    {
      "epoch": 6.73,
      "learning_rate": 1.9439421338155517e-06,
      "loss": 0.0003,
      "step": 5315
    },
    {
      "epoch": 6.73,
      "learning_rate": 1.9349005424954795e-06,
      "loss": 0.0005,
      "step": 5316
    },
    {
      "epoch": 6.73,
      "learning_rate": 1.925858951175407e-06,
      "loss": 0.2135,
      "step": 5317
    },
    {
      "epoch": 6.73,
      "learning_rate": 1.9168173598553347e-06,
      "loss": 0.0003,
      "step": 5318
    },
    {
      "epoch": 6.73,
      "learning_rate": 1.9077757685352625e-06,
      "loss": 0.0003,
      "step": 5319
    },
    {
      "epoch": 6.73,
      "learning_rate": 1.8987341772151901e-06,
      "loss": 0.0003,
      "step": 5320
    },
    {
      "epoch": 6.73,
      "learning_rate": 1.8896925858951177e-06,
      "loss": 0.0474,
      "step": 5321
    },
    {
      "epoch": 6.74,
      "learning_rate": 1.8806509945750453e-06,
      "loss": 0.0007,
      "step": 5322
    },
    {
      "epoch": 6.74,
      "learning_rate": 1.871609403254973e-06,
      "loss": 0.0394,
      "step": 5323
    },
    {
      "epoch": 6.74,
      "learning_rate": 1.8625678119349007e-06,
      "loss": 0.0006,
      "step": 5324
    },
    {
      "epoch": 6.74,
      "learning_rate": 1.8535262206148283e-06,
      "loss": 0.0178,
      "step": 5325
    },
    {
      "epoch": 6.74,
      "learning_rate": 1.844484629294756e-06,
      "loss": 0.0004,
      "step": 5326
    },
    {
      "epoch": 6.74,
      "learning_rate": 1.8354430379746838e-06,
      "loss": 0.0002,
      "step": 5327
    },
    {
      "epoch": 6.74,
      "learning_rate": 1.8264014466546114e-06,
      "loss": 0.0004,
      "step": 5328
    },
    {
      "epoch": 6.75,
      "learning_rate": 1.817359855334539e-06,
      "loss": 0.0006,
      "step": 5329
    },
    {
      "epoch": 6.75,
      "learning_rate": 1.8083182640144668e-06,
      "loss": 0.0003,
      "step": 5330
    },
    {
      "epoch": 6.75,
      "learning_rate": 1.7992766726943944e-06,
      "loss": 0.0347,
      "step": 5331
    },
    {
      "epoch": 6.75,
      "learning_rate": 1.790235081374322e-06,
      "loss": 0.0004,
      "step": 5332
    },
    {
      "epoch": 6.75,
      "learning_rate": 1.7811934900542498e-06,
      "loss": 0.0003,
      "step": 5333
    },
    {
      "epoch": 6.75,
      "learning_rate": 1.7721518987341774e-06,
      "loss": 0.0004,
      "step": 5334
    },
    {
      "epoch": 6.75,
      "learning_rate": 1.763110307414105e-06,
      "loss": 0.0003,
      "step": 5335
    },
    {
      "epoch": 6.75,
      "learning_rate": 1.7540687160940328e-06,
      "loss": 0.0005,
      "step": 5336
    },
    {
      "epoch": 6.76,
      "learning_rate": 1.7450271247739604e-06,
      "loss": 0.0002,
      "step": 5337
    },
    {
      "epoch": 6.76,
      "learning_rate": 1.735985533453888e-06,
      "loss": 0.0104,
      "step": 5338
    },
    {
      "epoch": 6.76,
      "learning_rate": 1.7269439421338156e-06,
      "loss": 0.001,
      "step": 5339
    },
    {
      "epoch": 6.76,
      "learning_rate": 1.7179023508137434e-06,
      "loss": 0.0003,
      "step": 5340
    },
    {
      "epoch": 6.76,
      "learning_rate": 1.708860759493671e-06,
      "loss": 0.0003,
      "step": 5341
    },
    {
      "epoch": 6.76,
      "learning_rate": 1.6998191681735986e-06,
      "loss": 0.0004,
      "step": 5342
    },
    {
      "epoch": 6.76,
      "learning_rate": 1.6907775768535265e-06,
      "loss": 0.0606,
      "step": 5343
    },
    {
      "epoch": 6.76,
      "learning_rate": 1.681735985533454e-06,
      "loss": 0.0004,
      "step": 5344
    },
    {
      "epoch": 6.77,
      "learning_rate": 1.6726943942133817e-06,
      "loss": 0.0003,
      "step": 5345
    },
    {
      "epoch": 6.77,
      "learning_rate": 1.6636528028933095e-06,
      "loss": 0.0004,
      "step": 5346
    },
    {
      "epoch": 6.77,
      "learning_rate": 1.654611211573237e-06,
      "loss": 0.0004,
      "step": 5347
    },
    {
      "epoch": 6.77,
      "learning_rate": 1.6455696202531647e-06,
      "loss": 0.0003,
      "step": 5348
    },
    {
      "epoch": 6.77,
      "learning_rate": 1.6365280289330925e-06,
      "loss": 0.0002,
      "step": 5349
    },
    {
      "epoch": 6.77,
      "learning_rate": 1.62748643761302e-06,
      "loss": 0.0366,
      "step": 5350
    },
    {
      "epoch": 6.77,
      "learning_rate": 1.6184448462929477e-06,
      "loss": 0.0003,
      "step": 5351
    },
    {
      "epoch": 6.77,
      "learning_rate": 1.6094032549728753e-06,
      "loss": 0.0003,
      "step": 5352
    },
    {
      "epoch": 6.78,
      "learning_rate": 1.6003616636528031e-06,
      "loss": 0.0125,
      "step": 5353
    },
    {
      "epoch": 6.78,
      "learning_rate": 1.5913200723327307e-06,
      "loss": 0.0009,
      "step": 5354
    },
    {
      "epoch": 6.78,
      "learning_rate": 1.5822784810126583e-06,
      "loss": 0.0848,
      "step": 5355
    },
    {
      "epoch": 6.78,
      "learning_rate": 1.5732368896925861e-06,
      "loss": 0.0003,
      "step": 5356
    },
    {
      "epoch": 6.78,
      "learning_rate": 1.5641952983725137e-06,
      "loss": 0.0017,
      "step": 5357
    },
    {
      "epoch": 6.78,
      "learning_rate": 1.5551537070524413e-06,
      "loss": 0.0004,
      "step": 5358
    },
    {
      "epoch": 6.78,
      "learning_rate": 1.546112115732369e-06,
      "loss": 0.0003,
      "step": 5359
    },
    {
      "epoch": 6.78,
      "learning_rate": 1.5370705244122965e-06,
      "loss": 0.0008,
      "step": 5360
    },
    {
      "epoch": 6.79,
      "learning_rate": 1.5280289330922242e-06,
      "loss": 0.0003,
      "step": 5361
    },
    {
      "epoch": 6.79,
      "learning_rate": 1.518987341772152e-06,
      "loss": 0.0002,
      "step": 5362
    },
    {
      "epoch": 6.79,
      "learning_rate": 1.5099457504520796e-06,
      "loss": 0.0014,
      "step": 5363
    },
    {
      "epoch": 6.79,
      "learning_rate": 1.5009041591320072e-06,
      "loss": 0.0003,
      "step": 5364
    },
    {
      "epoch": 6.79,
      "learning_rate": 1.491862567811935e-06,
      "loss": 0.0011,
      "step": 5365
    },
    {
      "epoch": 6.79,
      "learning_rate": 1.4828209764918626e-06,
      "loss": 0.0131,
      "step": 5366
    },
    {
      "epoch": 6.79,
      "learning_rate": 1.4737793851717902e-06,
      "loss": 0.0005,
      "step": 5367
    },
    {
      "epoch": 6.79,
      "learning_rate": 1.464737793851718e-06,
      "loss": 0.0004,
      "step": 5368
    },
    {
      "epoch": 6.8,
      "learning_rate": 1.4556962025316456e-06,
      "loss": 0.0003,
      "step": 5369
    },
    {
      "epoch": 6.8,
      "learning_rate": 1.4466546112115732e-06,
      "loss": 0.0003,
      "step": 5370
    },
    {
      "epoch": 6.8,
      "learning_rate": 1.437613019891501e-06,
      "loss": 0.0038,
      "step": 5371
    },
    {
      "epoch": 6.8,
      "learning_rate": 1.4285714285714286e-06,
      "loss": 0.0004,
      "step": 5372
    },
    {
      "epoch": 6.8,
      "learning_rate": 1.4195298372513562e-06,
      "loss": 0.0003,
      "step": 5373
    },
    {
      "epoch": 6.8,
      "learning_rate": 1.4104882459312838e-06,
      "loss": 0.0002,
      "step": 5374
    },
    {
      "epoch": 6.8,
      "learning_rate": 1.4014466546112116e-06,
      "loss": 0.0013,
      "step": 5375
    },
    {
      "epoch": 6.8,
      "learning_rate": 1.3924050632911392e-06,
      "loss": 0.0003,
      "step": 5376
    },
    {
      "epoch": 6.81,
      "learning_rate": 1.3833634719710669e-06,
      "loss": 0.0009,
      "step": 5377
    },
    {
      "epoch": 6.81,
      "learning_rate": 1.3743218806509947e-06,
      "loss": 0.0004,
      "step": 5378
    },
    {
      "epoch": 6.81,
      "learning_rate": 1.3652802893309223e-06,
      "loss": 0.0595,
      "step": 5379
    },
    {
      "epoch": 6.81,
      "learning_rate": 1.3562386980108499e-06,
      "loss": 0.0002,
      "step": 5380
    },
    {
      "epoch": 6.81,
      "learning_rate": 1.3471971066907777e-06,
      "loss": 0.0007,
      "step": 5381
    },
    {
      "epoch": 6.81,
      "learning_rate": 1.3381555153707053e-06,
      "loss": 0.0042,
      "step": 5382
    },
    {
      "epoch": 6.81,
      "learning_rate": 1.3291139240506329e-06,
      "loss": 0.0004,
      "step": 5383
    },
    {
      "epoch": 6.81,
      "learning_rate": 1.3200723327305607e-06,
      "loss": 0.0003,
      "step": 5384
    },
    {
      "epoch": 6.82,
      "learning_rate": 1.3110307414104883e-06,
      "loss": 0.0004,
      "step": 5385
    },
    {
      "epoch": 6.82,
      "learning_rate": 1.301989150090416e-06,
      "loss": 0.0942,
      "step": 5386
    },
    {
      "epoch": 6.82,
      "learning_rate": 1.2929475587703435e-06,
      "loss": 0.0025,
      "step": 5387
    },
    {
      "epoch": 6.82,
      "learning_rate": 1.2839059674502713e-06,
      "loss": 0.0003,
      "step": 5388
    },
    {
      "epoch": 6.82,
      "learning_rate": 1.274864376130199e-06,
      "loss": 0.0002,
      "step": 5389
    },
    {
      "epoch": 6.82,
      "learning_rate": 1.2658227848101265e-06,
      "loss": 0.0003,
      "step": 5390
    },
    {
      "epoch": 6.82,
      "learning_rate": 1.2567811934900543e-06,
      "loss": 0.0003,
      "step": 5391
    },
    {
      "epoch": 6.82,
      "learning_rate": 1.247739602169982e-06,
      "loss": 0.1813,
      "step": 5392
    },
    {
      "epoch": 6.83,
      "learning_rate": 1.2386980108499096e-06,
      "loss": 0.0005,
      "step": 5393
    },
    {
      "epoch": 6.83,
      "learning_rate": 1.2296564195298374e-06,
      "loss": 0.0006,
      "step": 5394
    },
    {
      "epoch": 6.83,
      "learning_rate": 1.220614828209765e-06,
      "loss": 0.0006,
      "step": 5395
    },
    {
      "epoch": 6.83,
      "learning_rate": 1.2115732368896926e-06,
      "loss": 0.0004,
      "step": 5396
    },
    {
      "epoch": 6.83,
      "learning_rate": 1.2025316455696204e-06,
      "loss": 0.0005,
      "step": 5397
    },
    {
      "epoch": 6.83,
      "learning_rate": 1.193490054249548e-06,
      "loss": 0.0046,
      "step": 5398
    },
    {
      "epoch": 6.83,
      "learning_rate": 1.1844484629294756e-06,
      "loss": 0.0009,
      "step": 5399
    },
    {
      "epoch": 6.83,
      "learning_rate": 1.1754068716094034e-06,
      "loss": 0.0085,
      "step": 5400
    },
    {
      "epoch": 6.84,
      "learning_rate": 1.166365280289331e-06,
      "loss": 0.0003,
      "step": 5401
    },
    {
      "epoch": 6.84,
      "learning_rate": 1.1573236889692586e-06,
      "loss": 0.0003,
      "step": 5402
    },
    {
      "epoch": 6.84,
      "learning_rate": 1.1482820976491862e-06,
      "loss": 0.0004,
      "step": 5403
    },
    {
      "epoch": 6.84,
      "learning_rate": 1.139240506329114e-06,
      "loss": 0.0009,
      "step": 5404
    },
    {
      "epoch": 6.84,
      "learning_rate": 1.1301989150090416e-06,
      "loss": 0.0003,
      "step": 5405
    },
    {
      "epoch": 6.84,
      "learning_rate": 1.1211573236889692e-06,
      "loss": 0.0006,
      "step": 5406
    },
    {
      "epoch": 6.84,
      "learning_rate": 1.112115732368897e-06,
      "loss": 0.0003,
      "step": 5407
    },
    {
      "epoch": 6.85,
      "learning_rate": 1.1030741410488247e-06,
      "loss": 0.0007,
      "step": 5408
    },
    {
      "epoch": 6.85,
      "learning_rate": 1.0940325497287523e-06,
      "loss": 0.004,
      "step": 5409
    },
    {
      "epoch": 6.85,
      "learning_rate": 1.08499095840868e-06,
      "loss": 0.0002,
      "step": 5410
    },
    {
      "epoch": 6.85,
      "learning_rate": 1.0759493670886077e-06,
      "loss": 0.0688,
      "step": 5411
    },
    {
      "epoch": 6.85,
      "learning_rate": 1.0669077757685353e-06,
      "loss": 0.0002,
      "step": 5412
    },
    {
      "epoch": 6.85,
      "learning_rate": 1.057866184448463e-06,
      "loss": 0.0002,
      "step": 5413
    },
    {
      "epoch": 6.85,
      "learning_rate": 1.0488245931283907e-06,
      "loss": 0.0005,
      "step": 5414
    },
    {
      "epoch": 6.85,
      "learning_rate": 1.0397830018083183e-06,
      "loss": 0.0004,
      "step": 5415
    },
    {
      "epoch": 6.86,
      "learning_rate": 1.030741410488246e-06,
      "loss": 0.0004,
      "step": 5416
    },
    {
      "epoch": 6.86,
      "learning_rate": 1.0216998191681737e-06,
      "loss": 0.0003,
      "step": 5417
    },
    {
      "epoch": 6.86,
      "learning_rate": 1.0126582278481013e-06,
      "loss": 0.0004,
      "step": 5418
    },
    {
      "epoch": 6.86,
      "learning_rate": 1.003616636528029e-06,
      "loss": 0.2085,
      "step": 5419
    },
    {
      "epoch": 6.86,
      "learning_rate": 9.945750452079567e-07,
      "loss": 0.0004,
      "step": 5420
    },
    {
      "epoch": 6.86,
      "learning_rate": 9.855334538878843e-07,
      "loss": 0.0004,
      "step": 5421
    },
    {
      "epoch": 6.86,
      "learning_rate": 9.76491862567812e-07,
      "loss": 0.0003,
      "step": 5422
    },
    {
      "epoch": 6.86,
      "learning_rate": 9.674502712477397e-07,
      "loss": 0.0003,
      "step": 5423
    },
    {
      "epoch": 6.87,
      "learning_rate": 9.584086799276674e-07,
      "loss": 0.0107,
      "step": 5424
    },
    {
      "epoch": 6.87,
      "learning_rate": 9.493670886075951e-07,
      "loss": 0.0007,
      "step": 5425
    },
    {
      "epoch": 6.87,
      "learning_rate": 9.403254972875227e-07,
      "loss": 0.1076,
      "step": 5426
    },
    {
      "epoch": 6.87,
      "learning_rate": 9.312839059674504e-07,
      "loss": 0.0026,
      "step": 5427
    },
    {
      "epoch": 6.87,
      "learning_rate": 9.22242314647378e-07,
      "loss": 0.0002,
      "step": 5428
    },
    {
      "epoch": 6.87,
      "learning_rate": 9.132007233273057e-07,
      "loss": 0.0941,
      "step": 5429
    },
    {
      "epoch": 6.87,
      "learning_rate": 9.041591320072334e-07,
      "loss": 0.0003,
      "step": 5430
    },
    {
      "epoch": 6.87,
      "learning_rate": 8.95117540687161e-07,
      "loss": 0.0002,
      "step": 5431
    },
    {
      "epoch": 6.88,
      "learning_rate": 8.860759493670887e-07,
      "loss": 0.0005,
      "step": 5432
    },
    {
      "epoch": 6.88,
      "learning_rate": 8.770343580470164e-07,
      "loss": 0.0355,
      "step": 5433
    },
    {
      "epoch": 6.88,
      "learning_rate": 8.67992766726944e-07,
      "loss": 0.0003,
      "step": 5434
    },
    {
      "epoch": 6.88,
      "learning_rate": 8.589511754068717e-07,
      "loss": 0.0002,
      "step": 5435
    },
    {
      "epoch": 6.88,
      "learning_rate": 8.499095840867993e-07,
      "loss": 0.0004,
      "step": 5436
    },
    {
      "epoch": 6.88,
      "learning_rate": 8.40867992766727e-07,
      "loss": 0.0005,
      "step": 5437
    },
    {
      "epoch": 6.88,
      "learning_rate": 8.318264014466547e-07,
      "loss": 0.0004,
      "step": 5438
    },
    {
      "epoch": 6.88,
      "learning_rate": 8.227848101265823e-07,
      "loss": 0.0003,
      "step": 5439
    },
    {
      "epoch": 6.89,
      "learning_rate": 8.1374321880651e-07,
      "loss": 0.0002,
      "step": 5440
    },
    {
      "epoch": 6.89,
      "learning_rate": 8.047016274864377e-07,
      "loss": 0.0003,
      "step": 5441
    },
    {
      "epoch": 6.89,
      "learning_rate": 7.956600361663654e-07,
      "loss": 0.0003,
      "step": 5442
    },
    {
      "epoch": 6.89,
      "learning_rate": 7.866184448462931e-07,
      "loss": 0.0006,
      "step": 5443
    },
    {
      "epoch": 6.89,
      "learning_rate": 7.775768535262207e-07,
      "loss": 0.0003,
      "step": 5444
    },
    {
      "epoch": 6.89,
      "learning_rate": 7.685352622061483e-07,
      "loss": 0.0586,
      "step": 5445
    },
    {
      "epoch": 6.89,
      "learning_rate": 7.59493670886076e-07,
      "loss": 0.0011,
      "step": 5446
    },
    {
      "epoch": 6.89,
      "learning_rate": 7.504520795660036e-07,
      "loss": 0.0007,
      "step": 5447
    },
    {
      "epoch": 6.9,
      "learning_rate": 7.414104882459313e-07,
      "loss": 0.0046,
      "step": 5448
    },
    {
      "epoch": 6.9,
      "learning_rate": 7.32368896925859e-07,
      "loss": 0.0002,
      "step": 5449
    },
    {
      "epoch": 6.9,
      "learning_rate": 7.233273056057866e-07,
      "loss": 0.0004,
      "step": 5450
    },
    {
      "epoch": 6.9,
      "learning_rate": 7.142857142857143e-07,
      "loss": 0.0002,
      "step": 5451
    },
    {
      "epoch": 6.9,
      "learning_rate": 7.052441229656419e-07,
      "loss": 0.0002,
      "step": 5452
    },
    {
      "epoch": 6.9,
      "learning_rate": 6.962025316455696e-07,
      "loss": 0.0002,
      "step": 5453
    },
    {
      "epoch": 6.9,
      "learning_rate": 6.871609403254973e-07,
      "loss": 0.0006,
      "step": 5454
    },
    {
      "epoch": 6.9,
      "learning_rate": 6.781193490054249e-07,
      "loss": 0.0005,
      "step": 5455
    },
    {
      "epoch": 6.91,
      "learning_rate": 6.690777576853526e-07,
      "loss": 0.0002,
      "step": 5456
    },
    {
      "epoch": 6.91,
      "learning_rate": 6.600361663652804e-07,
      "loss": 0.0002,
      "step": 5457
    },
    {
      "epoch": 6.91,
      "learning_rate": 6.50994575045208e-07,
      "loss": 0.0002,
      "step": 5458
    },
    {
      "epoch": 6.91,
      "learning_rate": 6.419529837251357e-07,
      "loss": 0.0004,
      "step": 5459
    },
    {
      "epoch": 6.91,
      "learning_rate": 6.329113924050633e-07,
      "loss": 0.0003,
      "step": 5460
    },
    {
      "epoch": 6.91,
      "learning_rate": 6.23869801084991e-07,
      "loss": 0.0003,
      "step": 5461
    },
    {
      "epoch": 6.91,
      "learning_rate": 6.148282097649187e-07,
      "loss": 0.0003,
      "step": 5462
    },
    {
      "epoch": 6.91,
      "learning_rate": 6.057866184448463e-07,
      "loss": 0.0004,
      "step": 5463
    },
    {
      "epoch": 6.92,
      "learning_rate": 5.96745027124774e-07,
      "loss": 0.001,
      "step": 5464
    },
    {
      "epoch": 6.92,
      "learning_rate": 5.877034358047017e-07,
      "loss": 0.0004,
      "step": 5465
    },
    {
      "epoch": 6.92,
      "learning_rate": 5.786618444846293e-07,
      "loss": 0.0038,
      "step": 5466
    },
    {
      "epoch": 6.92,
      "learning_rate": 5.69620253164557e-07,
      "loss": 0.0003,
      "step": 5467
    },
    {
      "epoch": 6.92,
      "learning_rate": 5.605786618444846e-07,
      "loss": 0.001,
      "step": 5468
    },
    {
      "epoch": 6.92,
      "learning_rate": 5.515370705244123e-07,
      "loss": 0.0004,
      "step": 5469
    },
    {
      "epoch": 6.92,
      "learning_rate": 5.4249547920434e-07,
      "loss": 0.071,
      "step": 5470
    },
    {
      "epoch": 6.92,
      "learning_rate": 5.334538878842676e-07,
      "loss": 0.0003,
      "step": 5471
    },
    {
      "epoch": 6.93,
      "learning_rate": 5.244122965641953e-07,
      "loss": 0.0003,
      "step": 5472
    },
    {
      "epoch": 6.93,
      "learning_rate": 5.15370705244123e-07,
      "loss": 0.0006,
      "step": 5473
    },
    {
      "epoch": 6.93,
      "learning_rate": 5.063291139240507e-07,
      "loss": 0.0628,
      "step": 5474
    },
    {
      "epoch": 6.93,
      "learning_rate": 4.972875226039784e-07,
      "loss": 0.0005,
      "step": 5475
    },
    {
      "epoch": 6.93,
      "learning_rate": 4.88245931283906e-07,
      "loss": 0.0009,
      "step": 5476
    },
    {
      "epoch": 6.93,
      "learning_rate": 4.792043399638337e-07,
      "loss": 0.0005,
      "step": 5477
    },
    {
      "epoch": 6.93,
      "learning_rate": 4.7016274864376133e-07,
      "loss": 0.0004,
      "step": 5478
    },
    {
      "epoch": 6.93,
      "learning_rate": 4.61121157323689e-07,
      "loss": 0.0004,
      "step": 5479
    },
    {
      "epoch": 6.94,
      "learning_rate": 4.520795660036167e-07,
      "loss": 0.0005,
      "step": 5480
    },
    {
      "epoch": 6.94,
      "learning_rate": 4.4303797468354435e-07,
      "loss": 0.0003,
      "step": 5481
    },
    {
      "epoch": 6.94,
      "learning_rate": 4.33996383363472e-07,
      "loss": 0.0683,
      "step": 5482
    },
    {
      "epoch": 6.94,
      "learning_rate": 4.2495479204339966e-07,
      "loss": 0.0004,
      "step": 5483
    },
    {
      "epoch": 6.94,
      "learning_rate": 4.1591320072332737e-07,
      "loss": 0.0008,
      "step": 5484
    },
    {
      "epoch": 6.94,
      "learning_rate": 4.06871609403255e-07,
      "loss": 0.0006,
      "step": 5485
    },
    {
      "epoch": 6.94,
      "learning_rate": 3.978300180831827e-07,
      "loss": 0.0002,
      "step": 5486
    },
    {
      "epoch": 6.94,
      "learning_rate": 3.8878842676311034e-07,
      "loss": 0.0013,
      "step": 5487
    },
    {
      "epoch": 6.95,
      "learning_rate": 3.79746835443038e-07,
      "loss": 0.0003,
      "step": 5488
    },
    {
      "epoch": 6.95,
      "learning_rate": 3.7070524412296565e-07,
      "loss": 0.0003,
      "step": 5489
    },
    {
      "epoch": 6.95,
      "learning_rate": 3.616636528028933e-07,
      "loss": 0.0002,
      "step": 5490
    },
    {
      "epoch": 6.95,
      "learning_rate": 3.5262206148282096e-07,
      "loss": 0.0004,
      "step": 5491
    },
    {
      "epoch": 6.95,
      "learning_rate": 3.4358047016274867e-07,
      "loss": 0.0101,
      "step": 5492
    },
    {
      "epoch": 6.95,
      "learning_rate": 3.345388788426763e-07,
      "loss": 0.0013,
      "step": 5493
    },
    {
      "epoch": 6.95,
      "learning_rate": 3.25497287522604e-07,
      "loss": 0.0003,
      "step": 5494
    },
    {
      "epoch": 6.96,
      "learning_rate": 3.1645569620253163e-07,
      "loss": 0.0004,
      "step": 5495
    },
    {
      "epoch": 6.96,
      "learning_rate": 3.0741410488245934e-07,
      "loss": 0.0002,
      "step": 5496
    },
    {
      "epoch": 6.96,
      "learning_rate": 2.98372513562387e-07,
      "loss": 0.001,
      "step": 5497
    },
    {
      "epoch": 6.96,
      "learning_rate": 2.8933092224231465e-07,
      "loss": 0.0013,
      "step": 5498
    },
    {
      "epoch": 6.96,
      "learning_rate": 2.802893309222423e-07,
      "loss": 0.0006,
      "step": 5499
    },
    {
      "epoch": 6.96,
      "learning_rate": 2.7124773960217e-07,
      "loss": 0.0004,
      "step": 5500
    },
    {
      "epoch": 6.96,
      "eval_accuracy": 0.9054054054054054,
      "eval_loss": 0.4979327917098999,
      "eval_runtime": 45.1232,
      "eval_samples_per_second": 29.519,
      "eval_steps_per_second": 3.701,
      "step": 5500
    }
  ],
  "max_steps": 5530,
  "num_train_epochs": 7,
  "total_flos": 3.622515599767142e+16,
  "trial_name": null,
  "trial_params": null
}