{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 5.97092084006462,
  "global_step": 462,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.01,
      "learning_rate": 1.4285714285714286e-06,
      "loss": 2.1703,
      "step": 1
    },
    {
      "epoch": 0.03,
      "learning_rate": 2.8571428571428573e-06,
      "loss": 2.1529,
      "step": 2
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.2857142857142855e-06,
      "loss": 1.4343,
      "step": 3
    },
    {
      "epoch": 0.05,
      "learning_rate": 5.7142857142857145e-06,
      "loss": 1.3822,
      "step": 4
    },
    {
      "epoch": 0.06,
      "learning_rate": 7.1428571428571436e-06,
      "loss": 1.259,
      "step": 5
    },
    {
      "epoch": 0.08,
      "learning_rate": 8.571428571428571e-06,
      "loss": 1.231,
      "step": 6
    },
    {
      "epoch": 0.09,
      "learning_rate": 1e-05,
      "loss": 1.1535,
      "step": 7
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.1428571428571429e-05,
      "loss": 1.1701,
      "step": 8
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.2857142857142859e-05,
      "loss": 1.1408,
      "step": 9
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.4285714285714287e-05,
      "loss": 1.1468,
      "step": 10
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.5714285714285715e-05,
      "loss": 1.1389,
      "step": 11
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.7142857142857142e-05,
      "loss": 1.0866,
      "step": 12
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.8571428571428575e-05,
      "loss": 1.0794,
      "step": 13
    },
    {
      "epoch": 0.18,
      "learning_rate": 2e-05,
      "loss": 1.0941,
      "step": 14
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.9999754126376247e-05,
      "loss": 1.0402,
      "step": 15
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.9999016517595752e-05,
      "loss": 1.0839,
      "step": 16
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.9997787209930222e-05,
      "loss": 1.0533,
      "step": 17
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.9996066263830533e-05,
      "loss": 1.0485,
      "step": 18
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.9993853763923724e-05,
      "loss": 1.0572,
      "step": 19
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.999114981900887e-05,
      "loss": 1.0493,
      "step": 20
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.9987954562051724e-05,
      "loss": 1.077,
      "step": 21
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.998426815017817e-05,
      "loss": 1.0676,
      "step": 22
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.9980090764666486e-05,
      "loss": 1.0695,
      "step": 23
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.9975422610938463e-05,
      "loss": 1.0383,
      "step": 24
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.9970263918549274e-05,
      "loss": 1.0036,
      "step": 25
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.9964614941176194e-05,
      "loss": 1.0124,
      "step": 26
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.9958475956606133e-05,
      "loss": 1.0127,
      "step": 27
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.995184726672197e-05,
      "loss": 0.9904,
      "step": 28
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.9944729197487702e-05,
      "loss": 1.0295,
      "step": 29
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.9937122098932428e-05,
      "loss": 1.0293,
      "step": 30
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.992902634513312e-05,
      "loss": 0.9826,
      "step": 31
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.9920442334196248e-05,
      "loss": 0.9924,
      "step": 32
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.9911370488238185e-05,
      "loss": 0.9813,
      "step": 33
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.9901811253364458e-05,
      "loss": 0.9965,
      "step": 34
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.989176509964781e-05,
      "loss": 1.0157,
      "step": 35
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.988123252110509e-05,
      "loss": 1.0625,
      "step": 36
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.9870214035672945e-05,
      "loss": 1.0077,
      "step": 37
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.985871018518236e-05,
      "loss": 0.9727,
      "step": 38
    },
    {
      "epoch": 0.5,
      "learning_rate": 1.984672153533202e-05,
      "loss": 0.9996,
      "step": 39
    },
    {
      "epoch": 0.52,
      "learning_rate": 1.9834248675660484e-05,
      "loss": 1.0046,
      "step": 40
    },
    {
      "epoch": 0.53,
      "learning_rate": 1.982129221951719e-05,
      "loss": 0.9891,
      "step": 41
    },
    {
      "epoch": 0.54,
      "learning_rate": 1.9807852804032306e-05,
      "loss": 1.1151,
      "step": 42
    },
    {
      "epoch": 0.56,
      "learning_rate": 1.9793931090085385e-05,
      "loss": 0.9895,
      "step": 43
    },
    {
      "epoch": 0.57,
      "learning_rate": 1.9779527762272877e-05,
      "loss": 0.9969,
      "step": 44
    },
    {
      "epoch": 0.58,
      "learning_rate": 1.976464352887447e-05,
      "loss": 0.9717,
      "step": 45
    },
    {
      "epoch": 0.59,
      "learning_rate": 1.9749279121818235e-05,
      "loss": 1.0397,
      "step": 46
    },
    {
      "epoch": 0.61,
      "learning_rate": 1.973343529664467e-05,
      "loss": 0.978,
      "step": 47
    },
    {
      "epoch": 0.62,
      "learning_rate": 1.971711283246951e-05,
      "loss": 0.9927,
      "step": 48
    },
    {
      "epoch": 0.63,
      "learning_rate": 1.9700312531945444e-05,
      "loss": 0.9446,
      "step": 49
    },
    {
      "epoch": 0.65,
      "learning_rate": 1.9683035221222617e-05,
      "loss": 0.9793,
      "step": 50
    },
    {
      "epoch": 0.66,
      "learning_rate": 1.9665281749908034e-05,
      "loss": 0.9999,
      "step": 51
    },
    {
      "epoch": 0.67,
      "learning_rate": 1.964705299102376e-05,
      "loss": 0.9607,
      "step": 52
    },
    {
      "epoch": 0.68,
      "learning_rate": 1.9628349840963997e-05,
      "loss": 0.9746,
      "step": 53
    },
    {
      "epoch": 0.7,
      "learning_rate": 1.9609173219450998e-05,
      "loss": 0.9671,
      "step": 54
    },
    {
      "epoch": 0.71,
      "learning_rate": 1.958952406948985e-05,
      "loss": 0.9879,
      "step": 55
    },
    {
      "epoch": 0.72,
      "learning_rate": 1.956940335732209e-05,
      "loss": 0.982,
      "step": 56
    },
    {
      "epoch": 0.74,
      "learning_rate": 1.9548812072378208e-05,
      "loss": 0.945,
      "step": 57
    },
    {
      "epoch": 0.75,
      "learning_rate": 1.9527751227228964e-05,
      "loss": 0.9689,
      "step": 58
    },
    {
      "epoch": 0.76,
      "learning_rate": 1.950622185753563e-05,
      "loss": 0.9955,
      "step": 59
    },
    {
      "epoch": 0.78,
      "learning_rate": 1.9484225021999032e-05,
      "loss": 0.9633,
      "step": 60
    },
    {
      "epoch": 0.79,
      "learning_rate": 1.9461761802307494e-05,
      "loss": 0.966,
      "step": 61
    },
    {
      "epoch": 0.8,
      "learning_rate": 1.9438833303083677e-05,
      "loss": 0.9962,
      "step": 62
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.941544065183021e-05,
      "loss": 1.0164,
      "step": 63
    },
    {
      "epoch": 0.83,
      "learning_rate": 1.939158499887428e-05,
      "loss": 0.9462,
      "step": 64
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.9367267517311057e-05,
      "loss": 1.0057,
      "step": 65
    },
    {
      "epoch": 0.85,
      "learning_rate": 1.9342489402945997e-05,
      "loss": 0.9601,
      "step": 66
    },
    {
      "epoch": 0.87,
      "learning_rate": 1.9317251874236066e-05,
      "loss": 0.9497,
      "step": 67
    },
    {
      "epoch": 0.88,
      "learning_rate": 1.9291556172229784e-05,
      "loss": 1.0107,
      "step": 68
    },
    {
      "epoch": 0.89,
      "learning_rate": 1.9265403560506223e-05,
      "loss": 0.951,
      "step": 69
    },
    {
      "epoch": 0.9,
      "learning_rate": 1.9238795325112867e-05,
      "loss": 0.9485,
      "step": 70
    },
    {
      "epoch": 0.92,
      "learning_rate": 1.9211732774502372e-05,
      "loss": 0.9925,
      "step": 71
    },
    {
      "epoch": 0.93,
      "learning_rate": 1.9184217239468213e-05,
      "loss": 0.9399,
      "step": 72
    },
    {
      "epoch": 0.94,
      "learning_rate": 1.915625007307925e-05,
      "loss": 0.9579,
      "step": 73
    },
    {
      "epoch": 0.96,
      "learning_rate": 1.912783265061319e-05,
      "loss": 0.9431,
      "step": 74
    },
    {
      "epoch": 0.97,
      "learning_rate": 1.9098966369488967e-05,
      "loss": 0.9758,
      "step": 75
    },
    {
      "epoch": 0.98,
      "learning_rate": 1.9069652649198004e-05,
      "loss": 0.9403,
      "step": 76
    },
    {
      "epoch": 1.0,
      "learning_rate": 1.9039892931234434e-05,
      "loss": 0.9434,
      "step": 77
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.900968867902419e-05,
      "loss": 0.9028,
      "step": 78
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.8979041377853068e-05,
      "loss": 0.8579,
      "step": 79
    },
    {
      "epoch": 1.03,
      "learning_rate": 1.8947952534793663e-05,
      "loss": 0.8575,
      "step": 80
    },
    {
      "epoch": 1.05,
      "learning_rate": 1.891642367863127e-05,
      "loss": 0.8453,
      "step": 81
    },
    {
      "epoch": 1.06,
      "learning_rate": 1.8884456359788725e-05,
      "loss": 0.8504,
      "step": 82
    },
    {
      "epoch": 1.07,
      "learning_rate": 1.8852052150250123e-05,
      "loss": 0.8479,
      "step": 83
    },
    {
      "epoch": 1.09,
      "learning_rate": 1.881921264348355e-05,
      "loss": 0.8473,
      "step": 84
    },
    {
      "epoch": 1.1,
      "learning_rate": 1.878593945436272e-05,
      "loss": 0.8223,
      "step": 85
    },
    {
      "epoch": 1.11,
      "learning_rate": 1.8752234219087538e-05,
      "loss": 0.8424,
      "step": 86
    },
    {
      "epoch": 1.12,
      "learning_rate": 1.871809859510368e-05,
      "loss": 0.8525,
      "step": 87
    },
    {
      "epoch": 1.14,
      "learning_rate": 1.8683534261021058e-05,
      "loss": 0.8561,
      "step": 88
    },
    {
      "epoch": 1.15,
      "learning_rate": 1.8648542916531283e-05,
      "loss": 0.8301,
      "step": 89
    },
    {
      "epoch": 1.16,
      "learning_rate": 1.8613126282324092e-05,
      "loss": 0.8174,
      "step": 90
    },
    {
      "epoch": 1.18,
      "learning_rate": 1.8577286100002723e-05,
      "loss": 0.8228,
      "step": 91
    },
    {
      "epoch": 1.19,
      "learning_rate": 1.8541024131998277e-05,
      "loss": 0.8189,
      "step": 92
    },
    {
      "epoch": 1.2,
      "learning_rate": 1.850434216148305e-05,
      "loss": 0.804,
      "step": 93
    },
    {
      "epoch": 1.21,
      "learning_rate": 1.8467241992282842e-05,
      "loss": 0.8368,
      "step": 94
    },
    {
      "epoch": 1.23,
      "learning_rate": 1.8429725448788267e-05,
      "loss": 0.8353,
      "step": 95
    },
    {
      "epoch": 1.24,
      "learning_rate": 1.8391794375865025e-05,
      "loss": 0.8153,
      "step": 96
    },
    {
      "epoch": 1.25,
      "learning_rate": 1.8353450638763178e-05,
      "loss": 0.778,
      "step": 97
    },
    {
      "epoch": 1.27,
      "learning_rate": 1.8314696123025456e-05,
      "loss": 0.8143,
      "step": 98
    },
    {
      "epoch": 1.28,
      "learning_rate": 1.827553273439449e-05,
      "loss": 0.8126,
      "step": 99
    },
    {
      "epoch": 1.29,
      "learning_rate": 1.823596239871915e-05,
      "loss": 0.7918,
      "step": 100
    },
    {
      "epoch": 1.31,
      "learning_rate": 1.819598706185979e-05,
      "loss": 0.8084,
      "step": 101
    },
    {
      "epoch": 1.32,
      "learning_rate": 1.8155608689592604e-05,
      "loss": 0.8166,
      "step": 102
    },
    {
      "epoch": 1.33,
      "learning_rate": 1.811482926751293e-05,
      "loss": 0.8211,
      "step": 103
    },
    {
      "epoch": 1.34,
      "learning_rate": 1.8073650800937627e-05,
      "loss": 0.83,
      "step": 104
    },
    {
      "epoch": 1.36,
      "learning_rate": 1.803207531480645e-05,
      "loss": 0.8166,
      "step": 105
    },
    {
      "epoch": 1.37,
      "learning_rate": 1.7990104853582494e-05,
      "loss": 0.8037,
      "step": 106
    },
    {
      "epoch": 1.38,
      "learning_rate": 1.7947741481151628e-05,
      "loss": 0.7819,
      "step": 107
    },
    {
      "epoch": 1.4,
      "learning_rate": 1.7904987280721037e-05,
      "loss": 0.781,
      "step": 108
    },
    {
      "epoch": 1.41,
      "learning_rate": 1.7861844354716757e-05,
      "loss": 0.8338,
      "step": 109
    },
    {
      "epoch": 1.42,
      "learning_rate": 1.78183148246803e-05,
      "loss": 0.8088,
      "step": 110
    },
    {
      "epoch": 1.43,
      "learning_rate": 1.777440083116432e-05,
      "loss": 0.7566,
      "step": 111
    },
    {
      "epoch": 1.45,
      "learning_rate": 1.773010453362737e-05,
      "loss": 0.786,
      "step": 112
    },
    {
      "epoch": 1.46,
      "learning_rate": 1.7685428110327683e-05,
      "loss": 0.787,
      "step": 113
    },
    {
      "epoch": 1.47,
      "learning_rate": 1.7640373758216075e-05,
      "loss": 0.7889,
      "step": 114
    },
    {
      "epoch": 1.49,
      "learning_rate": 1.7594943692827913e-05,
      "loss": 0.7934,
      "step": 115
    },
    {
      "epoch": 1.5,
      "learning_rate": 1.754914014817416e-05,
      "loss": 0.7853,
      "step": 116
    },
    {
      "epoch": 1.51,
      "learning_rate": 1.7502965376631515e-05,
      "loss": 0.807,
      "step": 117
    },
    {
      "epoch": 1.53,
      "learning_rate": 1.7456421648831658e-05,
      "loss": 0.7973,
      "step": 118
    },
    {
      "epoch": 1.54,
      "learning_rate": 1.7409511253549592e-05,
      "loss": 0.8228,
      "step": 119
    },
    {
      "epoch": 1.55,
      "learning_rate": 1.7362236497591097e-05,
      "loss": 0.7885,
      "step": 120
    },
    {
      "epoch": 1.56,
      "learning_rate": 1.731459970567928e-05,
      "loss": 0.7657,
      "step": 121
    },
    {
      "epoch": 1.58,
      "learning_rate": 1.7266603220340273e-05,
      "loss": 0.7701,
      "step": 122
    },
    {
      "epoch": 1.59,
      "learning_rate": 1.7218249401788033e-05,
      "loss": 0.7784,
      "step": 123
    },
    {
      "epoch": 1.6,
      "learning_rate": 1.7169540627808276e-05,
      "loss": 0.8019,
      "step": 124
    },
    {
      "epoch": 1.62,
      "learning_rate": 1.7120479293641558e-05,
      "loss": 0.7428,
      "step": 125
    },
    {
      "epoch": 1.63,
      "learning_rate": 1.7071067811865477e-05,
      "loss": 0.7482,
      "step": 126
    },
    {
      "epoch": 1.64,
      "learning_rate": 1.7021308612276056e-05,
      "loss": 0.7833,
      "step": 127
    },
    {
      "epoch": 1.65,
      "learning_rate": 1.6971204141768235e-05,
      "loss": 0.7865,
      "step": 128
    },
    {
      "epoch": 1.67,
      "learning_rate": 1.6920756864215558e-05,
      "loss": 0.7722,
      "step": 129
    },
    {
      "epoch": 1.68,
      "learning_rate": 1.686996926034902e-05,
      "loss": 0.7703,
      "step": 130
    },
    {
      "epoch": 1.69,
      "learning_rate": 1.6818843827635052e-05,
      "loss": 0.7667,
      "step": 131
    },
    {
      "epoch": 1.71,
      "learning_rate": 1.6767383080152744e-05,
      "loss": 0.7361,
      "step": 132
    },
    {
      "epoch": 1.72,
      "learning_rate": 1.6715589548470187e-05,
      "loss": 0.8037,
      "step": 133
    },
    {
      "epoch": 1.73,
      "learning_rate": 1.6663465779520042e-05,
      "loss": 0.7281,
      "step": 134
    },
    {
      "epoch": 1.74,
      "learning_rate": 1.6611014336474303e-05,
      "loss": 0.7402,
      "step": 135
    },
    {
      "epoch": 1.76,
      "learning_rate": 1.6558237798618243e-05,
      "loss": 0.7343,
      "step": 136
    },
    {
      "epoch": 1.77,
      "learning_rate": 1.6505138761223586e-05,
      "loss": 0.7379,
      "step": 137
    },
    {
      "epoch": 1.78,
      "learning_rate": 1.645171983542088e-05,
      "loss": 0.7296,
      "step": 138
    },
    {
      "epoch": 1.8,
      "learning_rate": 1.6397983648071093e-05,
      "loss": 0.7136,
      "step": 139
    },
    {
      "epoch": 1.81,
      "learning_rate": 1.6343932841636455e-05,
      "loss": 0.7394,
      "step": 140
    },
    {
      "epoch": 1.82,
      "learning_rate": 1.6289570074050492e-05,
      "loss": 0.7298,
      "step": 141
    },
    {
      "epoch": 1.84,
      "learning_rate": 1.6234898018587336e-05,
      "loss": 0.7253,
      "step": 142
    },
    {
      "epoch": 1.85,
      "learning_rate": 1.617991936373027e-05,
      "loss": 0.7967,
      "step": 143
    },
    {
      "epoch": 1.86,
      "learning_rate": 1.6124636813039502e-05,
      "loss": 0.6995,
      "step": 144
    },
    {
      "epoch": 1.87,
      "learning_rate": 1.6069053085019258e-05,
      "loss": 0.7193,
      "step": 145
    },
    {
      "epoch": 1.89,
      "learning_rate": 1.601317091298406e-05,
      "loss": 0.7227,
      "step": 146
    },
    {
      "epoch": 1.9,
      "learning_rate": 1.5956993044924334e-05,
      "loss": 0.7113,
      "step": 147
    },
    {
      "epoch": 1.91,
      "learning_rate": 1.5900522243371283e-05,
      "loss": 0.7434,
      "step": 148
    },
    {
      "epoch": 1.93,
      "learning_rate": 1.5843761285261027e-05,
      "loss": 0.7252,
      "step": 149
    },
    {
      "epoch": 1.94,
      "learning_rate": 1.578671296179806e-05,
      "loss": 0.7414,
      "step": 150
    },
    {
      "epoch": 1.95,
      "learning_rate": 1.5729380078317982e-05,
      "loss": 0.6879,
      "step": 151
    },
    {
      "epoch": 1.96,
      "learning_rate": 1.5671765454149558e-05,
      "loss": 0.7181,
      "step": 152
    },
    {
      "epoch": 1.98,
      "learning_rate": 1.5613871922476082e-05,
      "loss": 0.6852,
      "step": 153
    },
    {
      "epoch": 1.99,
      "learning_rate": 1.5555702330196024e-05,
      "loss": 0.7135,
      "step": 154
    },
    {
      "epoch": 2.0,
      "learning_rate": 1.5497259537783084e-05,
      "loss": 0.682,
      "step": 155
    },
    {
      "epoch": 2.02,
      "learning_rate": 1.543854641914549e-05,
      "loss": 0.4981,
      "step": 156
    },
    {
      "epoch": 2.03,
      "learning_rate": 1.537956586148469e-05,
      "loss": 0.4879,
      "step": 157
    },
    {
      "epoch": 2.04,
      "learning_rate": 1.5320320765153367e-05,
      "loss": 0.4878,
      "step": 158
    },
    {
      "epoch": 2.05,
      "learning_rate": 1.5260814043512838e-05,
      "loss": 0.4754,
      "step": 159
    },
    {
      "epoch": 2.07,
      "learning_rate": 1.5201048622789747e-05,
      "loss": 0.4727,
      "step": 160
    },
    {
      "epoch": 2.08,
      "learning_rate": 1.5141027441932217e-05,
      "loss": 0.4714,
      "step": 161
    },
    {
      "epoch": 2.09,
      "learning_rate": 1.5080753452465296e-05,
      "loss": 0.4644,
      "step": 162
    },
    {
      "epoch": 2.11,
      "learning_rate": 1.502022961834582e-05,
      "loss": 0.4727,
      "step": 163
    },
    {
      "epoch": 2.12,
      "learning_rate": 1.4959458915816681e-05,
      "loss": 0.4385,
      "step": 164
    },
    {
      "epoch": 2.13,
      "learning_rate": 1.4898444333260436e-05,
      "loss": 0.4349,
      "step": 165
    },
    {
      "epoch": 2.15,
      "learning_rate": 1.4837188871052399e-05,
      "loss": 0.4708,
      "step": 166
    },
    {
      "epoch": 2.16,
      "learning_rate": 1.4775695541413063e-05,
      "loss": 0.435,
      "step": 167
    },
    {
      "epoch": 2.17,
      "learning_rate": 1.4713967368259981e-05,
      "loss": 0.4391,
      "step": 168
    },
    {
      "epoch": 2.18,
      "learning_rate": 1.4652007387059077e-05,
      "loss": 0.4417,
      "step": 169
    },
    {
      "epoch": 2.2,
      "learning_rate": 1.4589818644675378e-05,
      "loss": 0.4331,
      "step": 170
    },
    {
      "epoch": 2.21,
      "learning_rate": 1.4527404199223173e-05,
      "loss": 0.4214,
      "step": 171
    },
    {
      "epoch": 2.22,
      "learning_rate": 1.446476711991563e-05,
      "loss": 0.4067,
      "step": 172
    },
    {
      "epoch": 2.24,
      "learning_rate": 1.4401910486913892e-05,
      "loss": 0.4371,
      "step": 173
    },
    {
      "epoch": 2.25,
      "learning_rate": 1.4338837391175582e-05,
      "loss": 0.4318,
      "step": 174
    },
    {
      "epoch": 2.26,
      "learning_rate": 1.4275550934302822e-05,
      "loss": 0.4215,
      "step": 175
    },
    {
      "epoch": 2.27,
      "learning_rate": 1.4212054228389712e-05,
      "loss": 0.4312,
      "step": 176
    },
    {
      "epoch": 2.29,
      "learning_rate": 1.4148350395869279e-05,
      "loss": 0.4097,
      "step": 177
    },
    {
      "epoch": 2.3,
      "learning_rate": 1.4084442569359964e-05,
      "loss": 0.4094,
      "step": 178
    },
    {
      "epoch": 2.31,
      "learning_rate": 1.4020333891511536e-05,
      "loss": 0.4075,
      "step": 179
    },
    {
      "epoch": 2.33,
      "learning_rate": 1.395602751485059e-05,
      "loss": 0.4258,
      "step": 180
    },
    {
      "epoch": 2.34,
      "learning_rate": 1.3891526601625492e-05,
      "loss": 0.3956,
      "step": 181
    },
    {
      "epoch": 2.35,
      "learning_rate": 1.3826834323650899e-05,
      "loss": 0.3987,
      "step": 182
    },
    {
      "epoch": 2.37,
      "learning_rate": 1.3761953862151773e-05,
      "loss": 0.3987,
      "step": 183
    },
    {
      "epoch": 2.38,
      "learning_rate": 1.3696888407606952e-05,
      "loss": 0.3704,
      "step": 184
    },
    {
      "epoch": 2.39,
      "learning_rate": 1.3631641159592253e-05,
      "loss": 0.4014,
      "step": 185
    },
    {
      "epoch": 2.4,
      "learning_rate": 1.3566215326623131e-05,
      "loss": 0.4135,
      "step": 186
    },
    {
      "epoch": 2.42,
      "learning_rate": 1.3500614125996924e-05,
      "loss": 0.3753,
      "step": 187
    },
    {
      "epoch": 2.43,
      "learning_rate": 1.3434840783634611e-05,
      "loss": 0.3808,
      "step": 188
    },
    {
      "epoch": 2.44,
      "learning_rate": 1.3368898533922202e-05,
      "loss": 0.3618,
      "step": 189
    },
    {
      "epoch": 2.46,
      "learning_rate": 1.3302790619551673e-05,
      "loss": 0.3765,
      "step": 190
    },
    {
      "epoch": 2.47,
      "learning_rate": 1.3236520291361516e-05,
      "loss": 0.3512,
      "step": 191
    },
    {
      "epoch": 2.48,
      "learning_rate": 1.3170090808176883e-05,
      "loss": 0.3367,
      "step": 192
    },
    {
      "epoch": 2.49,
      "learning_rate": 1.310350543664932e-05,
      "loss": 0.3799,
      "step": 193
    },
    {
      "epoch": 2.51,
      "learning_rate": 1.3036767451096148e-05,
      "loss": 0.3589,
      "step": 194
    },
    {
      "epoch": 2.52,
      "learning_rate": 1.2969880133339437e-05,
      "loss": 0.3658,
      "step": 195
    },
    {
      "epoch": 2.53,
      "learning_rate": 1.2902846772544625e-05,
      "loss": 0.3567,
      "step": 196
    },
    {
      "epoch": 2.55,
      "learning_rate": 1.2835670665058779e-05,
      "loss": 0.3312,
      "step": 197
    },
    {
      "epoch": 2.56,
      "learning_rate": 1.2768355114248493e-05,
      "loss": 0.3518,
      "step": 198
    },
    {
      "epoch": 2.57,
      "learning_rate": 1.2700903430337456e-05,
      "loss": 0.3588,
      "step": 199
    },
    {
      "epoch": 2.58,
      "learning_rate": 1.2633318930243647e-05,
      "loss": 0.3709,
      "step": 200
    },
    {
      "epoch": 2.6,
      "learning_rate": 1.2565604937416267e-05,
      "loss": 0.3446,
      "step": 201
    },
    {
      "epoch": 2.61,
      "learning_rate": 1.249776478167227e-05,
      "loss": 0.363,
      "step": 202
    },
    {
      "epoch": 2.62,
      "learning_rate": 1.242980179903264e-05,
      "loss": 0.3603,
      "step": 203
    },
    {
      "epoch": 2.64,
      "learning_rate": 1.2361719331558346e-05,
      "loss": 0.3517,
      "step": 204
    },
    {
      "epoch": 2.65,
      "learning_rate": 1.229352072718598e-05,
      "loss": 0.3329,
      "step": 205
    },
    {
      "epoch": 2.66,
      "learning_rate": 1.2225209339563144e-05,
      "loss": 0.3441,
      "step": 206
    },
    {
      "epoch": 2.68,
      "learning_rate": 1.2156788527883524e-05,
      "loss": 0.335,
      "step": 207
    },
    {
      "epoch": 2.69,
      "learning_rate": 1.20882616567217e-05,
      "loss": 0.3259,
      "step": 208
    },
    {
      "epoch": 2.7,
      "learning_rate": 1.2019632095867697e-05,
      "loss": 0.296,
      "step": 209
    },
    {
      "epoch": 2.71,
      "learning_rate": 1.1950903220161286e-05,
      "loss": 0.321,
      "step": 210
    },
    {
      "epoch": 2.73,
      "learning_rate": 1.1882078409326003e-05,
      "loss": 0.3059,
      "step": 211
    },
    {
      "epoch": 2.74,
      "learning_rate": 1.1813161047802986e-05,
      "loss": 0.2979,
      "step": 212
    },
    {
      "epoch": 2.75,
      "learning_rate": 1.174415452458451e-05,
      "loss": 0.281,
      "step": 213
    },
    {
      "epoch": 2.77,
      "learning_rate": 1.1675062233047365e-05,
      "loss": 0.3029,
      "step": 214
    },
    {
      "epoch": 2.78,
      "learning_rate": 1.1605887570785972e-05,
      "loss": 0.3045,
      "step": 215
    },
    {
      "epoch": 2.79,
      "learning_rate": 1.1536633939445302e-05,
      "loss": 0.2989,
      "step": 216
    },
    {
      "epoch": 2.8,
      "learning_rate": 1.1467304744553618e-05,
      "loss": 0.2804,
      "step": 217
    },
    {
      "epoch": 2.82,
      "learning_rate": 1.1397903395354996e-05,
      "loss": 0.2852,
      "step": 218
    },
    {
      "epoch": 2.83,
      "learning_rate": 1.132843330464168e-05,
      "loss": 0.248,
      "step": 219
    },
    {
      "epoch": 2.84,
      "learning_rate": 1.1258897888586256e-05,
      "loss": 0.2822,
      "step": 220
    },
    {
      "epoch": 2.86,
      "learning_rate": 1.118930056657367e-05,
      "loss": 0.2713,
      "step": 221
    },
    {
      "epoch": 2.87,
      "learning_rate": 1.1119644761033079e-05,
      "loss": 0.2879,
      "step": 222
    },
    {
      "epoch": 2.88,
      "learning_rate": 1.1049933897269547e-05,
      "loss": 0.257,
      "step": 223
    },
    {
      "epoch": 2.89,
      "learning_rate": 1.098017140329561e-05,
      "loss": 0.2643,
      "step": 224
    },
    {
      "epoch": 2.91,
      "learning_rate": 1.0910360709662701e-05,
      "loss": 0.2694,
      "step": 225
    },
    {
      "epoch": 2.92,
      "learning_rate": 1.0840505249292477e-05,
      "loss": 0.2558,
      "step": 226
    },
    {
      "epoch": 2.93,
      "learning_rate": 1.0770608457307965e-05,
      "loss": 0.2547,
      "step": 227
    },
    {
      "epoch": 2.95,
      "learning_rate": 1.0700673770864673e-05,
      "loss": 0.2287,
      "step": 228
    },
    {
      "epoch": 2.96,
      "learning_rate": 1.0630704628981561e-05,
      "loss": 0.2441,
      "step": 229
    },
    {
      "epoch": 2.97,
      "learning_rate": 1.0560704472371919e-05,
      "loss": 0.2165,
      "step": 230
    },
    {
      "epoch": 2.99,
      "learning_rate": 1.0490676743274181e-05,
      "loss": 0.2302,
      "step": 231
    },
    {
      "epoch": 3.0,
      "learning_rate": 1.0420624885282653e-05,
      "loss": 0.2407,
      "step": 232
    },
    {
      "epoch": 3.01,
      "learning_rate": 1.0350552343178164e-05,
      "loss": 0.1206,
      "step": 233
    },
    {
      "epoch": 3.02,
      "learning_rate": 1.028046256275869e-05,
      "loss": 0.0922,
      "step": 234
    },
    {
      "epoch": 3.04,
      "learning_rate": 1.0210358990669889e-05,
      "loss": 0.0915,
      "step": 235
    },
    {
      "epoch": 3.05,
      "learning_rate": 1.0140245074235624e-05,
      "loss": 0.0834,
      "step": 236
    },
    {
      "epoch": 3.06,
      "learning_rate": 1.0070124261288437e-05,
      "loss": 0.0781,
      "step": 237
    },
    {
      "epoch": 3.08,
      "learning_rate": 1e-05,
      "loss": 0.0863,
      "step": 238
    },
    {
      "epoch": 3.09,
      "learning_rate": 9.929875738711565e-06,
      "loss": 0.0917,
      "step": 239
    },
    {
      "epoch": 3.1,
      "learning_rate": 9.85975492576438e-06,
      "loss": 0.0788,
      "step": 240
    },
    {
      "epoch": 3.11,
      "learning_rate": 9.789641009330113e-06,
      "loss": 0.0799,
      "step": 241
    },
    {
      "epoch": 3.13,
      "learning_rate": 9.719537437241311e-06,
      "loss": 0.0847,
      "step": 242
    },
    {
      "epoch": 3.14,
      "learning_rate": 9.64944765682184e-06,
      "loss": 0.0869,
      "step": 243
    },
    {
      "epoch": 3.15,
      "learning_rate": 9.579375114717352e-06,
      "loss": 0.0804,
      "step": 244
    },
    {
      "epoch": 3.17,
      "learning_rate": 9.50932325672582e-06,
      "loss": 0.0854,
      "step": 245
    },
    {
      "epoch": 3.18,
      "learning_rate": 9.439295527628083e-06,
      "loss": 0.0715,
      "step": 246
    },
    {
      "epoch": 3.19,
      "learning_rate": 9.369295371018442e-06,
      "loss": 0.0814,
      "step": 247
    },
    {
      "epoch": 3.21,
      "learning_rate": 9.299326229135326e-06,
      "loss": 0.0745,
      "step": 248
    },
    {
      "epoch": 3.22,
      "learning_rate": 9.22939154269204e-06,
      "loss": 0.0758,
      "step": 249
    },
    {
      "epoch": 3.23,
      "learning_rate": 9.159494750707527e-06,
      "loss": 0.0762,
      "step": 250
    },
    {
      "epoch": 3.24,
      "learning_rate": 9.0896392903373e-06,
      "loss": 0.0713,
      "step": 251
    },
    {
      "epoch": 3.26,
      "learning_rate": 9.019828596704394e-06,
      "loss": 0.0831,
      "step": 252
    },
    {
      "epoch": 3.27,
      "learning_rate": 8.950066102730456e-06,
      "loss": 0.0747,
      "step": 253
    },
    {
      "epoch": 3.28,
      "learning_rate": 8.880355238966923e-06,
      "loss": 0.0968,
      "step": 254
    },
    {
      "epoch": 3.3,
      "learning_rate": 8.81069943342633e-06,
      "loss": 0.0591,
      "step": 255
    },
    {
      "epoch": 3.31,
      "learning_rate": 8.741102111413749e-06,
      "loss": 0.0755,
      "step": 256
    },
    {
      "epoch": 3.32,
      "learning_rate": 8.671566695358324e-06,
      "loss": 0.0697,
      "step": 257
    },
    {
      "epoch": 3.33,
      "learning_rate": 8.602096604645009e-06,
      "loss": 0.0691,
      "step": 258
    },
    {
      "epoch": 3.35,
      "learning_rate": 8.532695255446384e-06,
      "loss": 0.0739,
      "step": 259
    },
    {
      "epoch": 3.36,
      "learning_rate": 8.463366060554698e-06,
      "loss": 0.0676,
      "step": 260
    },
    {
      "epoch": 3.37,
      "learning_rate": 8.394112429214032e-06,
      "loss": 0.0683,
      "step": 261
    },
    {
      "epoch": 3.39,
      "learning_rate": 8.324937766952638e-06,
      "loss": 0.0696,
      "step": 262
    },
    {
      "epoch": 3.4,
      "learning_rate": 8.255845475415493e-06,
      "loss": 0.0665,
      "step": 263
    },
    {
      "epoch": 3.41,
      "learning_rate": 8.186838952197019e-06,
      "loss": 0.0722,
      "step": 264
    },
    {
      "epoch": 3.42,
      "learning_rate": 8.117921590674002e-06,
      "loss": 0.0699,
      "step": 265
    },
    {
      "epoch": 3.44,
      "learning_rate": 8.04909677983872e-06,
      "loss": 0.0661,
      "step": 266
    },
    {
      "epoch": 3.45,
      "learning_rate": 7.980367904132303e-06,
      "loss": 0.0688,
      "step": 267
    },
    {
      "epoch": 3.46,
      "learning_rate": 7.911738343278303e-06,
      "loss": 0.0634,
      "step": 268
    },
    {
      "epoch": 3.48,
      "learning_rate": 7.843211472116476e-06,
      "loss": 0.0631,
      "step": 269
    },
    {
      "epoch": 3.49,
      "learning_rate": 7.774790660436857e-06,
      "loss": 0.068,
      "step": 270
    },
    {
      "epoch": 3.5,
      "learning_rate": 7.706479272814024e-06,
      "loss": 0.0596,
      "step": 271
    },
    {
      "epoch": 3.52,
      "learning_rate": 7.63828066844166e-06,
      "loss": 0.0603,
      "step": 272
    },
    {
      "epoch": 3.53,
      "learning_rate": 7.570198200967363e-06,
      "loss": 0.0631,
      "step": 273
    },
    {
      "epoch": 3.54,
      "learning_rate": 7.50223521832773e-06,
      "loss": 0.0727,
      "step": 274
    },
    {
      "epoch": 3.55,
      "learning_rate": 7.434395062583735e-06,
      "loss": 0.0635,
      "step": 275
    },
    {
      "epoch": 3.57,
      "learning_rate": 7.366681069756352e-06,
      "loss": 0.0575,
      "step": 276
    },
    {
      "epoch": 3.58,
      "learning_rate": 7.299096569662549e-06,
      "loss": 0.0613,
      "step": 277
    },
    {
      "epoch": 3.59,
      "learning_rate": 7.2316448857515076e-06,
      "loss": 0.0593,
      "step": 278
    },
    {
      "epoch": 3.61,
      "learning_rate": 7.164329334941225e-06,
      "loss": 0.0557,
      "step": 279
    },
    {
      "epoch": 3.62,
      "learning_rate": 7.097153227455379e-06,
      "loss": 0.0572,
      "step": 280
    },
    {
      "epoch": 3.63,
      "learning_rate": 7.030119866660565e-06,
      "loss": 0.0551,
      "step": 281
    },
    {
      "epoch": 3.64,
      "learning_rate": 6.963232548903853e-06,
      "loss": 0.0548,
      "step": 282
    },
    {
      "epoch": 3.66,
      "learning_rate": 6.896494563350681e-06,
      "loss": 0.0565,
      "step": 283
    },
    {
      "epoch": 3.67,
      "learning_rate": 6.829909191823121e-06,
      "loss": 0.0519,
      "step": 284
    },
    {
      "epoch": 3.68,
      "learning_rate": 6.763479708638485e-06,
      "loss": 0.0522,
      "step": 285
    },
    {
      "epoch": 3.7,
      "learning_rate": 6.697209380448333e-06,
      "loss": 0.0622,
      "step": 286
    },
    {
      "epoch": 3.71,
      "learning_rate": 6.631101466077801e-06,
      "loss": 0.0516,
      "step": 287
    },
    {
      "epoch": 3.72,
      "learning_rate": 6.5651592163653885e-06,
      "loss": 0.0496,
      "step": 288
    },
    {
      "epoch": 3.74,
      "learning_rate": 6.499385874003077e-06,
      "loss": 0.051,
      "step": 289
    },
    {
      "epoch": 3.75,
      "learning_rate": 6.43378467337687e-06,
      "loss": 0.0483,
      "step": 290
    },
    {
      "epoch": 3.76,
      "learning_rate": 6.368358840407754e-06,
      "loss": 0.0395,
      "step": 291
    },
    {
      "epoch": 3.77,
      "learning_rate": 6.303111592393051e-06,
      "loss": 0.0453,
      "step": 292
    },
    {
      "epoch": 3.79,
      "learning_rate": 6.23804613784823e-06,
      "loss": 0.0423,
      "step": 293
    },
    {
      "epoch": 3.8,
      "learning_rate": 6.173165676349103e-06,
      "loss": 0.0416,
      "step": 294
    },
    {
      "epoch": 3.81,
      "learning_rate": 6.108473398374509e-06,
      "loss": 0.0402,
      "step": 295
    },
    {
      "epoch": 3.83,
      "learning_rate": 6.043972485149414e-06,
      "loss": 0.0418,
      "step": 296
    },
    {
      "epoch": 3.84,
      "learning_rate": 5.979666108488464e-06,
      "loss": 0.0432,
      "step": 297
    },
    {
      "epoch": 3.85,
      "learning_rate": 5.91555743064004e-06,
      "loss": 0.0354,
      "step": 298
    },
    {
      "epoch": 3.86,
      "learning_rate": 5.851649604130723e-06,
      "loss": 0.0353,
      "step": 299
    },
    {
      "epoch": 3.88,
      "learning_rate": 5.787945771610296e-06,
      "loss": 0.033,
      "step": 300
    },
    {
      "epoch": 3.89,
      "learning_rate": 5.724449065697182e-06,
      "loss": 0.0344,
      "step": 301
    },
    {
      "epoch": 3.9,
      "learning_rate": 5.66116260882442e-06,
      "loss": 0.0339,
      "step": 302
    },
    {
      "epoch": 3.92,
      "learning_rate": 5.598089513086108e-06,
      "loss": 0.034,
      "step": 303
    },
    {
      "epoch": 3.93,
      "learning_rate": 5.535232880084373e-06,
      "loss": 0.0348,
      "step": 304
    },
    {
      "epoch": 3.94,
      "learning_rate": 5.472595800776831e-06,
      "loss": 0.0345,
      "step": 305
    },
    {
      "epoch": 3.95,
      "learning_rate": 5.410181355324622e-06,
      "loss": 0.0312,
      "step": 306
    },
    {
      "epoch": 3.97,
      "learning_rate": 5.347992612940927e-06,
      "loss": 0.0336,
      "step": 307
    },
    {
      "epoch": 3.98,
      "learning_rate": 5.286032631740023e-06,
      "loss": 0.0369,
      "step": 308
    },
    {
      "epoch": 3.99,
      "learning_rate": 5.22430445858694e-06,
      "loss": 0.0285,
      "step": 309
    },
    {
      "epoch": 4.01,
      "learning_rate": 5.1628111289476025e-06,
      "loss": 0.0239,
      "step": 310
    },
    {
      "epoch": 4.02,
      "learning_rate": 5.101555666739563e-06,
      "loss": 0.0095,
      "step": 311
    },
    {
      "epoch": 4.03,
      "learning_rate": 5.040541084183326e-06,
      "loss": 0.0114,
      "step": 312
    },
    {
      "epoch": 4.05,
      "learning_rate": 4.979770381654181e-06,
      "loss": 0.0109,
      "step": 313
    },
    {
      "epoch": 4.06,
      "learning_rate": 4.919246547534709e-06,
      "loss": 0.0095,
      "step": 314
    },
    {
      "epoch": 4.07,
      "learning_rate": 4.858972558067784e-06,
      "loss": 0.0103,
      "step": 315
    },
    {
      "epoch": 4.08,
      "learning_rate": 4.798951377210253e-06,
      "loss": 0.0088,
      "step": 316
    },
    {
      "epoch": 4.1,
      "learning_rate": 4.739185956487169e-06,
      "loss": 0.0107,
      "step": 317
    },
    {
      "epoch": 4.11,
      "learning_rate": 4.679679234846636e-06,
      "loss": 0.0108,
      "step": 318
    },
    {
      "epoch": 4.12,
      "learning_rate": 4.6204341385153186e-06,
      "loss": 0.0098,
      "step": 319
    },
    {
      "epoch": 4.14,
      "learning_rate": 4.561453580854516e-06,
      "loss": 0.0121,
      "step": 320
    },
    {
      "epoch": 4.15,
      "learning_rate": 4.502740462216919e-06,
      "loss": 0.0124,
      "step": 321
    },
    {
      "epoch": 4.16,
      "learning_rate": 4.444297669803981e-06,
      "loss": 0.0089,
      "step": 322
    },
    {
      "epoch": 4.17,
      "learning_rate": 4.386128077523923e-06,
      "loss": 0.0119,
      "step": 323
    },
    {
      "epoch": 4.19,
      "learning_rate": 4.328234545850441e-06,
      "loss": 0.0124,
      "step": 324
    },
    {
      "epoch": 4.2,
      "learning_rate": 4.270619921682019e-06,
      "loss": 0.0098,
      "step": 325
    },
    {
      "epoch": 4.21,
      "learning_rate": 4.213287038201943e-06,
      "loss": 0.0106,
      "step": 326
    },
    {
      "epoch": 4.23,
      "learning_rate": 4.156238714738974e-06,
      "loss": 0.012,
      "step": 327
    },
    {
      "epoch": 4.24,
      "learning_rate": 4.09947775662872e-06,
      "loss": 0.0081,
      "step": 328
    },
    {
      "epoch": 4.25,
      "learning_rate": 4.043006955075667e-06,
      "loss": 0.0093,
      "step": 329
    },
    {
      "epoch": 4.26,
      "learning_rate": 3.986829087015941e-06,
      "loss": 0.0079,
      "step": 330
    },
    {
      "epoch": 4.28,
      "learning_rate": 3.930946914980744e-06,
      "loss": 0.0105,
      "step": 331
    },
    {
      "epoch": 4.29,
      "learning_rate": 3.875363186960499e-06,
      "loss": 0.0094,
      "step": 332
    },
    {
      "epoch": 4.3,
      "learning_rate": 3.820080636269737e-06,
      "loss": 0.0095,
      "step": 333
    },
    {
      "epoch": 4.32,
      "learning_rate": 3.7651019814126656e-06,
      "loss": 0.0105,
      "step": 334
    },
    {
      "epoch": 4.33,
      "learning_rate": 3.7104299259495113e-06,
      "loss": 0.0058,
      "step": 335
    },
    {
      "epoch": 4.34,
      "learning_rate": 3.6560671583635467e-06,
      "loss": 0.0071,
      "step": 336
    },
    {
      "epoch": 4.36,
      "learning_rate": 3.6020163519289077e-06,
      "loss": 0.009,
      "step": 337
    },
    {
      "epoch": 4.37,
      "learning_rate": 3.5482801645791266e-06,
      "loss": 0.0086,
      "step": 338
    },
    {
      "epoch": 4.38,
      "learning_rate": 3.494861238776418e-06,
      "loss": 0.0073,
      "step": 339
    },
    {
      "epoch": 4.39,
      "learning_rate": 3.4417622013817597e-06,
      "loss": 0.0073,
      "step": 340
    },
    {
      "epoch": 4.41,
      "learning_rate": 3.3889856635257024e-06,
      "loss": 0.0054,
      "step": 341
    },
    {
      "epoch": 4.42,
      "learning_rate": 3.3365342204799613e-06,
      "loss": 0.0061,
      "step": 342
    },
    {
      "epoch": 4.43,
      "learning_rate": 3.284410451529816e-06,
      "loss": 0.0055,
      "step": 343
    },
    {
      "epoch": 4.45,
      "learning_rate": 3.2326169198472555e-06,
      "loss": 0.0066,
      "step": 344
    },
    {
      "epoch": 4.46,
      "learning_rate": 3.1811561723649496e-06,
      "loss": 0.009,
      "step": 345
    },
    {
      "epoch": 4.47,
      "learning_rate": 3.1300307396509833e-06,
      "loss": 0.0077,
      "step": 346
    },
    {
      "epoch": 4.48,
      "learning_rate": 3.0792431357844444e-06,
      "loss": 0.0061,
      "step": 347
    },
    {
      "epoch": 4.5,
      "learning_rate": 3.028795858231768e-06,
      "loss": 0.0062,
      "step": 348
    },
    {
      "epoch": 4.51,
      "learning_rate": 2.9786913877239486e-06,
      "loss": 0.0058,
      "step": 349
    },
    {
      "epoch": 4.52,
      "learning_rate": 2.9289321881345257e-06,
      "loss": 0.0078,
      "step": 350
    },
    {
      "epoch": 4.54,
      "learning_rate": 2.879520706358446e-06,
      "loss": 0.0075,
      "step": 351
    },
    {
      "epoch": 4.55,
      "learning_rate": 2.8304593721917283e-06,
      "loss": 0.0058,
      "step": 352
    },
    {
      "epoch": 4.56,
      "learning_rate": 2.7817505982119708e-06,
      "loss": 0.0065,
      "step": 353
    },
    {
      "epoch": 4.58,
      "learning_rate": 2.7333967796597317e-06,
      "loss": 0.0058,
      "step": 354
    },
    {
      "epoch": 4.59,
      "learning_rate": 2.6854002943207245e-06,
      "loss": 0.0063,
      "step": 355
    },
    {
      "epoch": 4.6,
      "learning_rate": 2.637763502408909e-06,
      "loss": 0.0052,
      "step": 356
    },
    {
      "epoch": 4.61,
      "learning_rate": 2.5904887464504115e-06,
      "loss": 0.0068,
      "step": 357
    },
    {
      "epoch": 4.63,
      "learning_rate": 2.5435783511683444e-06,
      "loss": 0.0089,
      "step": 358
    },
    {
      "epoch": 4.64,
      "learning_rate": 2.4970346233684863e-06,
      "loss": 0.006,
      "step": 359
    },
    {
      "epoch": 4.65,
      "learning_rate": 2.450859851825842e-06,
      "loss": 0.0046,
      "step": 360
    },
    {
      "epoch": 4.67,
      "learning_rate": 2.4050563071720867e-06,
      "loss": 0.0055,
      "step": 361
    },
    {
      "epoch": 4.68,
      "learning_rate": 2.3596262417839256e-06,
      "loss": 0.0045,
      "step": 362
    },
    {
      "epoch": 4.69,
      "learning_rate": 2.3145718896723204e-06,
      "loss": 0.0053,
      "step": 363
    },
    {
      "epoch": 4.7,
      "learning_rate": 2.26989546637263e-06,
      "loss": 0.0056,
      "step": 364
    },
    {
      "epoch": 4.72,
      "learning_rate": 2.225599168835677e-06,
      "loss": 0.0043,
      "step": 365
    },
    {
      "epoch": 4.73,
      "learning_rate": 2.1816851753197023e-06,
      "loss": 0.0078,
      "step": 366
    },
    {
      "epoch": 4.74,
      "learning_rate": 2.138155645283244e-06,
      "loss": 0.0068,
      "step": 367
    },
    {
      "epoch": 4.76,
      "learning_rate": 2.095012719278966e-06,
      "loss": 0.0037,
      "step": 368
    },
    {
      "epoch": 4.77,
      "learning_rate": 2.0522585188483745e-06,
      "loss": 0.0038,
      "step": 369
    },
    {
      "epoch": 4.78,
      "learning_rate": 2.009895146417512e-06,
      "loss": 0.0044,
      "step": 370
    },
    {
      "epoch": 4.79,
      "learning_rate": 1.967924685193552e-06,
      "loss": 0.0059,
      "step": 371
    },
    {
      "epoch": 4.81,
      "learning_rate": 1.9263491990623763e-06,
      "loss": 0.005,
      "step": 372
    },
    {
      "epoch": 4.82,
      "learning_rate": 1.885170732487074e-06,
      "loss": 0.0041,
      "step": 373
    },
    {
      "epoch": 4.83,
      "learning_rate": 1.8443913104073984e-06,
      "loss": 0.004,
      "step": 374
    },
    {
      "epoch": 4.85,
      "learning_rate": 1.8040129381402137e-06,
      "loss": 0.0042,
      "step": 375
    },
    {
      "epoch": 4.86,
      "learning_rate": 1.7640376012808536e-06,
      "loss": 0.0036,
      "step": 376
    },
    {
      "epoch": 4.87,
      "learning_rate": 1.7244672656055105e-06,
      "loss": 0.0037,
      "step": 377
    },
    {
      "epoch": 4.89,
      "learning_rate": 1.6853038769745466e-06,
      "loss": 0.0043,
      "step": 378
    },
    {
      "epoch": 4.9,
      "learning_rate": 1.6465493612368233e-06,
      "loss": 0.0052,
      "step": 379
    },
    {
      "epoch": 4.91,
      "learning_rate": 1.6082056241349787e-06,
      "loss": 0.0037,
      "step": 380
    },
    {
      "epoch": 4.92,
      "learning_rate": 1.5702745512117323e-06,
      "loss": 0.0034,
      "step": 381
    },
    {
      "epoch": 4.94,
      "learning_rate": 1.5327580077171589e-06,
      "loss": 0.0044,
      "step": 382
    },
    {
      "epoch": 4.95,
      "learning_rate": 1.495657838516953e-06,
      "loss": 0.0029,
      "step": 383
    },
    {
      "epoch": 4.96,
      "learning_rate": 1.4589758680017263e-06,
      "loss": 0.0026,
      "step": 384
    },
    {
      "epoch": 4.98,
      "learning_rate": 1.4227138999972801e-06,
      "loss": 0.0036,
      "step": 385
    },
    {
      "epoch": 4.99,
      "learning_rate": 1.3868737176759105e-06,
      "loss": 0.0049,
      "step": 386
    },
    {
      "epoch": 5.0,
      "learning_rate": 1.3514570834687203e-06,
      "loss": 0.0031,
      "step": 387
    },
    {
      "epoch": 5.01,
      "learning_rate": 1.3164657389789459e-06,
      "loss": 0.0015,
      "step": 388
    },
    {
      "epoch": 5.03,
      "learning_rate": 1.281901404896323e-06,
      "loss": 0.0016,
      "step": 389
    },
    {
      "epoch": 5.04,
      "learning_rate": 1.2477657809124632e-06,
      "loss": 0.0023,
      "step": 390
    },
    {
      "epoch": 5.05,
      "learning_rate": 1.2140605456372856e-06,
      "loss": 0.0026,
      "step": 391
    },
    {
      "epoch": 5.07,
      "learning_rate": 1.1807873565164507e-06,
      "loss": 0.0018,
      "step": 392
    },
    {
      "epoch": 5.08,
      "learning_rate": 1.1479478497498796e-06,
      "loss": 0.0017,
      "step": 393
    },
    {
      "epoch": 5.09,
      "learning_rate": 1.1155436402112785e-06,
      "loss": 0.0021,
      "step": 394
    },
    {
      "epoch": 5.11,
      "learning_rate": 1.08357632136873e-06,
      "loss": 0.0014,
      "step": 395
    },
    {
      "epoch": 5.12,
      "learning_rate": 1.0520474652063395e-06,
      "loss": 0.0012,
      "step": 396
    },
    {
      "epoch": 5.13,
      "learning_rate": 1.0209586221469336e-06,
      "loss": 0.0019,
      "step": 397
    },
    {
      "epoch": 5.14,
      "learning_rate": 9.903113209758098e-07,
      "loss": 0.0014,
      "step": 398
    },
    {
      "epoch": 5.16,
      "learning_rate": 9.601070687655667e-07,
      "loss": 0.0024,
      "step": 399
    },
    {
      "epoch": 5.17,
      "learning_rate": 9.303473508019944e-07,
      "loss": 0.003,
      "step": 400
    },
    {
      "epoch": 5.18,
      "learning_rate": 9.010336305110345e-07,
      "loss": 0.0029,
      "step": 401
    },
    {
      "epoch": 5.2,
      "learning_rate": 8.721673493868111e-07,
      "loss": 0.0014,
      "step": 402
    },
    {
      "epoch": 5.21,
      "learning_rate": 8.437499269207538e-07,
      "loss": 0.0018,
      "step": 403
    },
    {
      "epoch": 5.22,
      "learning_rate": 8.157827605317892e-07,
      "loss": 0.0013,
      "step": 404
    },
    {
      "epoch": 5.23,
      "learning_rate": 7.882672254976298e-07,
      "loss": 0.0014,
      "step": 405
    },
    {
      "epoch": 5.25,
      "learning_rate": 7.612046748871327e-07,
      "loss": 0.0014,
      "step": 406
    },
    {
      "epoch": 5.26,
      "learning_rate": 7.345964394937788e-07,
      "loss": 0.002,
      "step": 407
    },
    {
      "epoch": 5.27,
      "learning_rate": 7.084438277702188e-07,
      "loss": 0.0021,
      "step": 408
    },
    {
      "epoch": 5.29,
      "learning_rate": 6.827481257639346e-07,
      "loss": 0.0014,
      "step": 409
    },
    {
      "epoch": 5.3,
      "learning_rate": 6.57510597054003e-07,
      "loss": 0.0012,
      "step": 410
    },
    {
      "epoch": 5.31,
      "learning_rate": 6.327324826889469e-07,
      "loss": 0.002,
      "step": 411
    },
    {
      "epoch": 5.32,
      "learning_rate": 6.084150011257239e-07,
      "loss": 0.0014,
      "step": 412
    },
    {
      "epoch": 5.34,
      "learning_rate": 5.845593481697931e-07,
      "loss": 0.0011,
      "step": 413
    },
    {
      "epoch": 5.35,
      "learning_rate": 5.611666969163243e-07,
      "loss": 0.0025,
      "step": 414
    },
    {
      "epoch": 5.36,
      "learning_rate": 5.382381976925044e-07,
      "loss": 0.0013,
      "step": 415
    },
    {
      "epoch": 5.38,
      "learning_rate": 5.157749780009735e-07,
      "loss": 0.0014,
      "step": 416
    },
    {
      "epoch": 5.39,
      "learning_rate": 4.937781424643728e-07,
      "loss": 0.0014,
      "step": 417
    },
    {
      "epoch": 5.4,
      "learning_rate": 4.7224877277103673e-07,
      "loss": 0.0012,
      "step": 418
    },
    {
      "epoch": 5.42,
      "learning_rate": 4.511879276217967e-07,
      "loss": 0.0014,
      "step": 419
    },
    {
      "epoch": 5.43,
      "learning_rate": 4.305966426779118e-07,
      "loss": 0.0016,
      "step": 420
    },
    {
      "epoch": 5.44,
      "learning_rate": 4.1047593051015245e-07,
      "loss": 0.0012,
      "step": 421
    },
    {
      "epoch": 5.45,
      "learning_rate": 3.908267805490051e-07,
      "loss": 0.0024,
      "step": 422
    },
    {
      "epoch": 5.47,
      "learning_rate": 3.7165015903600553e-07,
      "loss": 0.0012,
      "step": 423
    },
    {
      "epoch": 5.48,
      "learning_rate": 3.529470089762421e-07,
      "loss": 0.0014,
      "step": 424
    },
    {
      "epoch": 5.49,
      "learning_rate": 3.347182500919677e-07,
      "loss": 0.0014,
      "step": 425
    },
    {
      "epoch": 5.51,
      "learning_rate": 3.1696477877738664e-07,
      "loss": 0.0021,
      "step": 426
    },
    {
      "epoch": 5.52,
      "learning_rate": 2.996874680545603e-07,
      "loss": 0.002,
      "step": 427
    },
    {
      "epoch": 5.53,
      "learning_rate": 2.8288716753049007e-07,
      "loss": 0.0013,
      "step": 428
    },
    {
      "epoch": 5.54,
      "learning_rate": 2.665647033553309e-07,
      "loss": 0.0014,
      "step": 429
    },
    {
      "epoch": 5.56,
      "learning_rate": 2.507208781817638e-07,
      "loss": 0.0012,
      "step": 430
    },
    {
      "epoch": 5.57,
      "learning_rate": 2.3535647112553295e-07,
      "loss": 0.0018,
      "step": 431
    },
    {
      "epoch": 5.58,
      "learning_rate": 2.20472237727124e-07,
      "loss": 0.0014,
      "step": 432
    },
    {
      "epoch": 5.6,
      "learning_rate": 2.0606890991461737e-07,
      "loss": 0.0011,
      "step": 433
    },
    {
      "epoch": 5.61,
      "learning_rate": 1.921471959676957e-07,
      "loss": 0.0012,
      "step": 434
    },
    {
      "epoch": 5.62,
      "learning_rate": 1.787077804828097e-07,
      "loss": 0.0012,
      "step": 435
    },
    {
      "epoch": 5.63,
      "learning_rate": 1.657513243395159e-07,
      "loss": 0.0014,
      "step": 436
    },
    {
      "epoch": 5.65,
      "learning_rate": 1.5327846466797857e-07,
      "loss": 0.0013,
      "step": 437
    },
    {
      "epoch": 5.66,
      "learning_rate": 1.4128981481764115e-07,
      "loss": 0.0012,
      "step": 438
    },
    {
      "epoch": 5.67,
      "learning_rate": 1.2978596432705826e-07,
      "loss": 0.0013,
      "step": 439
    },
    {
      "epoch": 5.69,
      "learning_rate": 1.1876747889491225e-07,
      "loss": 0.0017,
      "step": 440
    },
    {
      "epoch": 5.7,
      "learning_rate": 1.0823490035218986e-07,
      "loss": 0.0014,
      "step": 441
    },
    {
      "epoch": 5.71,
      "learning_rate": 9.818874663554356e-08,
      "loss": 0.0021,
      "step": 442
    },
    {
      "epoch": 5.73,
      "learning_rate": 8.862951176181744e-08,
      "loss": 0.0011,
      "step": 443
    },
    {
      "epoch": 5.74,
      "learning_rate": 7.955766580375334e-08,
      "loss": 0.0013,
      "step": 444
    },
    {
      "epoch": 5.75,
      "learning_rate": 7.097365486688158e-08,
      "loss": 0.0014,
      "step": 445
    },
    {
      "epoch": 5.76,
      "learning_rate": 6.287790106757396e-08,
      "loss": 0.0012,
      "step": 446
    },
    {
      "epoch": 5.78,
      "learning_rate": 5.527080251229833e-08,
      "loss": 0.0013,
      "step": 447
    },
    {
      "epoch": 5.79,
      "learning_rate": 4.815273327803183e-08,
      "loss": 0.0013,
      "step": 448
    },
    {
      "epoch": 5.8,
      "learning_rate": 4.152404339386795e-08,
      "loss": 0.0023,
      "step": 449
    },
    {
      "epoch": 5.82,
      "learning_rate": 3.538505882380916e-08,
      "loss": 0.0014,
      "step": 450
    },
    {
      "epoch": 5.83,
      "learning_rate": 2.9736081450730813e-08,
      "loss": 0.0013,
      "step": 451
    },
    {
      "epoch": 5.84,
      "learning_rate": 2.4577389061539724e-08,
      "loss": 0.0015,
      "step": 452
    },
    {
      "epoch": 5.85,
      "learning_rate": 1.9909235333517296e-08,
      "loss": 0.0013,
      "step": 453
    },
    {
      "epoch": 5.87,
      "learning_rate": 1.5731849821833955e-08,
      "loss": 0.0012,
      "step": 454
    },
    {
      "epoch": 5.88,
      "learning_rate": 1.2045437948275952e-08,
      "loss": 0.0012,
      "step": 455
    },
    {
      "epoch": 5.89,
      "learning_rate": 8.850180991131219e-09,
      "loss": 0.0013,
      "step": 456
    },
    {
      "epoch": 5.91,
      "learning_rate": 6.146236076279843e-09,
      "loss": 0.0013,
      "step": 457
    },
    {
      "epoch": 5.92,
      "learning_rate": 3.933736169471347e-09,
      "loss": 0.0019,
      "step": 458
    },
    {
      "epoch": 5.93,
      "learning_rate": 2.2127900697777038e-09,
      "loss": 0.0012,
      "step": 459
    },
    {
      "epoch": 5.95,
      "learning_rate": 9.83482404249836e-10,
      "loss": 0.0011,
      "step": 460
    },
    {
      "epoch": 5.96,
      "learning_rate": 2.45873623754278e-10,
      "loss": 0.0014,
      "step": 461
    },
    {
      "epoch": 5.97,
      "learning_rate": 0.0,
      "loss": 0.0012,
      "step": 462
    },
    {
      "epoch": 5.97,
      "step": 462,
      "total_flos": 1.223725601702019e+18,
      "train_loss": 0.3808029658807184,
      "train_runtime": 10388.2603,
      "train_samples_per_second": 11.435,
      "train_steps_per_second": 0.044
    }
  ],
  "max_steps": 462,
  "num_train_epochs": 6,
  "total_flos": 1.223725601702019e+18,
  "trial_name": null,
  "trial_params": null
}