{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.9875910971045894,
  "global_step": 474,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.01,
      "learning_rate": 1.3333333333333334e-06,
      "loss": 0.9833,
      "step": 1
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.666666666666667e-06,
      "loss": 0.9773,
      "step": 2
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.000000000000001e-06,
      "loss": 0.9057,
      "step": 3
    },
    {
      "epoch": 0.03,
      "learning_rate": 5.333333333333334e-06,
      "loss": 0.8608,
      "step": 4
    },
    {
      "epoch": 0.03,
      "learning_rate": 6.666666666666667e-06,
      "loss": 0.8369,
      "step": 5
    },
    {
      "epoch": 0.04,
      "learning_rate": 8.000000000000001e-06,
      "loss": 0.8389,
      "step": 6
    },
    {
      "epoch": 0.04,
      "learning_rate": 9.333333333333334e-06,
      "loss": 0.8248,
      "step": 7
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.0666666666666667e-05,
      "loss": 0.7964,
      "step": 8
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.2e-05,
      "loss": 0.7733,
      "step": 9
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.3333333333333333e-05,
      "loss": 0.754,
      "step": 10
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.4666666666666666e-05,
      "loss": 0.7854,
      "step": 11
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.6000000000000003e-05,
      "loss": 0.7464,
      "step": 12
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.7333333333333336e-05,
      "loss": 0.8009,
      "step": 13
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.866666666666667e-05,
      "loss": 0.7194,
      "step": 14
    },
    {
      "epoch": 0.09,
      "learning_rate": 2e-05,
      "loss": 0.7622,
      "step": 15
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.9999765769911108e-05,
      "loss": 0.7234,
      "step": 16
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.999906309061717e-05,
      "loss": 0.748,
      "step": 17
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.9997891995035914e-05,
      "loss": 0.7251,
      "step": 18
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.999625253802851e-05,
      "loss": 0.7541,
      "step": 19
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.9994144796396985e-05,
      "loss": 0.7397,
      "step": 20
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.999156886888064e-05,
      "loss": 0.7587,
      "step": 21
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.9988524876151425e-05,
      "loss": 0.7024,
      "step": 22
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.9985012960808275e-05,
      "loss": 0.7247,
      "step": 23
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.9981033287370443e-05,
      "loss": 0.7367,
      "step": 24
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.9976586042269776e-05,
      "loss": 0.671,
      "step": 25
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.9971671433842e-05,
      "loss": 0.699,
      "step": 26
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.9966289692316944e-05,
      "loss": 0.7455,
      "step": 27
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.9960441069807778e-05,
      "loss": 0.7342,
      "step": 28
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.9954125840299165e-05,
      "loss": 0.7215,
      "step": 29
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.9947344299634464e-05,
      "loss": 0.7298,
      "step": 30
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.994009676550185e-05,
      "loss": 0.731,
      "step": 31
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.9932383577419432e-05,
      "loss": 0.7422,
      "step": 32
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.992420509671936e-05,
      "loss": 0.7201,
      "step": 33
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.9915561706530882e-05,
      "loss": 0.6377,
      "step": 34
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.9906453811762415e-05,
      "loss": 0.7684,
      "step": 35
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.9896881839082554e-05,
      "loss": 0.6847,
      "step": 36
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.9886846236900102e-05,
      "loss": 0.7413,
      "step": 37
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.9876347475343062e-05,
      "loss": 0.7418,
      "step": 38
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.9865386046236597e-05,
      "loss": 0.7133,
      "step": 39
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.9853962463080013e-05,
      "loss": 0.7368,
      "step": 40
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.984207726102269e-05,
      "loss": 0.6957,
      "step": 41
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.982973099683902e-05,
      "loss": 0.7068,
      "step": 42
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.9816924248902304e-05,
      "loss": 0.7005,
      "step": 43
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.9803657617157693e-05,
      "loss": 0.7019,
      "step": 44
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.9789931723094046e-05,
      "loss": 0.8071,
      "step": 45
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.9775747209714847e-05,
      "loss": 0.6871,
      "step": 46
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.976110474150806e-05,
      "loss": 0.695,
      "step": 47
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.9746005004415004e-05,
      "loss": 0.6839,
      "step": 48
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.973044870579824e-05,
      "loss": 0.7331,
      "step": 49
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.9714436574408408e-05,
      "loss": 0.6614,
      "step": 50
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.9697969360350098e-05,
      "loss": 0.6847,
      "step": 51
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.9681047835046708e-05,
      "loss": 0.7143,
      "step": 52
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.9663672791204328e-05,
      "loss": 0.6828,
      "step": 53
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.9645845042774555e-05,
      "loss": 0.7189,
      "step": 54
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.962756542491641e-05,
      "loss": 0.7008,
      "step": 55
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.960883479395721e-05,
      "loss": 0.782,
      "step": 56
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.9589654027352412e-05,
      "loss": 0.6962,
      "step": 57
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.957002402364456e-05,
      "loss": 0.7084,
      "step": 58
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.9549945702421144e-05,
      "loss": 0.7008,
      "step": 59
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.9529420004271568e-05,
      "loss": 0.7173,
      "step": 60
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.950844789074305e-05,
      "loss": 0.678,
      "step": 61
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.9487030344295586e-05,
      "loss": 0.691,
      "step": 62
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.9465168368255946e-05,
      "loss": 0.6422,
      "step": 63
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.9442862986770645e-05,
      "loss": 0.7058,
      "step": 64
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.9420115244757985e-05,
      "loss": 0.7368,
      "step": 65
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.9396926207859085e-05,
      "loss": 0.6976,
      "step": 66
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.9373296962387988e-05,
      "loss": 0.7015,
      "step": 67
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.9349228615280736e-05,
      "loss": 0.6702,
      "step": 68
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.932472229404356e-05,
      "loss": 0.7055,
      "step": 69
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.929977914670001e-05,
      "loss": 0.7297,
      "step": 70
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.9274400341737214e-05,
      "loss": 0.6833,
      "step": 71
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.924858706805112e-05,
      "loss": 0.7191,
      "step": 72
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.9222340534890803e-05,
      "loss": 0.7267,
      "step": 73
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.9195661971801825e-05,
      "loss": 0.6683,
      "step": 74
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.9168552628568632e-05,
      "loss": 0.7173,
      "step": 75
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.914101377515599e-05,
      "loss": 0.6817,
      "step": 76
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.9113046701649517e-05,
      "loss": 0.6894,
      "step": 77
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.9084652718195237e-05,
      "loss": 0.676,
      "step": 78
    },
    {
      "epoch": 0.5,
      "learning_rate": 1.9055833154938208e-05,
      "loss": 0.6968,
      "step": 79
    },
    {
      "epoch": 0.5,
      "learning_rate": 1.90265893619602e-05,
      "loss": 0.6614,
      "step": 80
    },
    {
      "epoch": 0.51,
      "learning_rate": 1.8996922709216456e-05,
      "loss": 0.6928,
      "step": 81
    },
    {
      "epoch": 0.52,
      "learning_rate": 1.8966834586471517e-05,
      "loss": 0.6435,
      "step": 82
    },
    {
      "epoch": 0.52,
      "learning_rate": 1.8936326403234125e-05,
      "loss": 0.6777,
      "step": 83
    },
    {
      "epoch": 0.53,
      "learning_rate": 1.8905399588691165e-05,
      "loss": 0.6872,
      "step": 84
    },
    {
      "epoch": 0.54,
      "learning_rate": 1.8874055591640746e-05,
      "loss": 0.6696,
      "step": 85
    },
    {
      "epoch": 0.54,
      "learning_rate": 1.8842295880424305e-05,
      "loss": 0.6764,
      "step": 86
    },
    {
      "epoch": 0.55,
      "learning_rate": 1.8810121942857848e-05,
      "loss": 0.6735,
      "step": 87
    },
    {
      "epoch": 0.55,
      "learning_rate": 1.8777535286162217e-05,
      "loss": 0.6839,
      "step": 88
    },
    {
      "epoch": 0.56,
      "learning_rate": 1.8744537436892517e-05,
      "loss": 0.6723,
      "step": 89
    },
    {
      "epoch": 0.57,
      "learning_rate": 1.8711129940866577e-05,
      "loss": 0.7009,
      "step": 90
    },
    {
      "epoch": 0.57,
      "learning_rate": 1.8677314363092555e-05,
      "loss": 0.6782,
      "step": 91
    },
    {
      "epoch": 0.58,
      "learning_rate": 1.8643092287695604e-05,
      "loss": 0.6812,
      "step": 92
    },
    {
      "epoch": 0.59,
      "learning_rate": 1.860846531784368e-05,
      "loss": 0.7392,
      "step": 93
    },
    {
      "epoch": 0.59,
      "learning_rate": 1.8573435075672422e-05,
      "loss": 0.6991,
      "step": 94
    },
    {
      "epoch": 0.6,
      "learning_rate": 1.8538003202209186e-05,
      "loss": 0.6849,
      "step": 95
    },
    {
      "epoch": 0.61,
      "learning_rate": 1.8502171357296144e-05,
      "loss": 0.6697,
      "step": 96
    },
    {
      "epoch": 0.61,
      "learning_rate": 1.8465941219512533e-05,
      "loss": 0.686,
      "step": 97
    },
    {
      "epoch": 0.62,
      "learning_rate": 1.8429314486096042e-05,
      "loss": 0.6728,
      "step": 98
    },
    {
      "epoch": 0.62,
      "learning_rate": 1.839229287286327e-05,
      "loss": 0.7075,
      "step": 99
    },
    {
      "epoch": 0.63,
      "learning_rate": 1.8354878114129368e-05,
      "loss": 0.6484,
      "step": 100
    },
    {
      "epoch": 0.64,
      "learning_rate": 1.831707196262679e-05,
      "loss": 0.6735,
      "step": 101
    },
    {
      "epoch": 0.64,
      "learning_rate": 1.827887618942318e-05,
      "loss": 0.6843,
      "step": 102
    },
    {
      "epoch": 0.65,
      "learning_rate": 1.824029258383841e-05,
      "loss": 0.6252,
      "step": 103
    },
    {
      "epoch": 0.66,
      "learning_rate": 1.8201322953360758e-05,
      "loss": 0.6535,
      "step": 104
    },
    {
      "epoch": 0.66,
      "learning_rate": 1.816196912356222e-05,
      "loss": 0.6906,
      "step": 105
    },
    {
      "epoch": 0.67,
      "learning_rate": 1.812223293801301e-05,
      "loss": 0.6769,
      "step": 106
    },
    {
      "epoch": 0.67,
      "learning_rate": 1.8082116258195173e-05,
      "loss": 0.7302,
      "step": 107
    },
    {
      "epoch": 0.68,
      "learning_rate": 1.8041620963415418e-05,
      "loss": 0.6505,
      "step": 108
    },
    {
      "epoch": 0.69,
      "learning_rate": 1.800074895071704e-05,
      "loss": 0.6494,
      "step": 109
    },
    {
      "epoch": 0.69,
      "learning_rate": 1.795950213479107e-05,
      "loss": 0.7324,
      "step": 110
    },
    {
      "epoch": 0.7,
      "learning_rate": 1.7917882447886585e-05,
      "loss": 0.7598,
      "step": 111
    },
    {
      "epoch": 0.71,
      "learning_rate": 1.787589183972017e-05,
      "loss": 0.6745,
      "step": 112
    },
    {
      "epoch": 0.71,
      "learning_rate": 1.7833532277384607e-05,
      "loss": 0.6611,
      "step": 113
    },
    {
      "epoch": 0.72,
      "learning_rate": 1.7790805745256703e-05,
      "loss": 0.6535,
      "step": 114
    },
    {
      "epoch": 0.72,
      "learning_rate": 1.7747714244904348e-05,
      "loss": 0.6724,
      "step": 115
    },
    {
      "epoch": 0.73,
      "learning_rate": 1.7704259794992734e-05,
      "loss": 0.6335,
      "step": 116
    },
    {
      "epoch": 0.74,
      "learning_rate": 1.766044443118978e-05,
      "loss": 0.7166,
      "step": 117
    },
    {
      "epoch": 0.74,
      "learning_rate": 1.7616270206070814e-05,
      "loss": 0.6555,
      "step": 118
    },
    {
      "epoch": 0.75,
      "learning_rate": 1.7571739189022365e-05,
      "loss": 0.6502,
      "step": 119
    },
    {
      "epoch": 0.76,
      "learning_rate": 1.7526853466145248e-05,
      "loss": 0.6635,
      "step": 120
    },
    {
      "epoch": 0.76,
      "learning_rate": 1.7481615140156837e-05,
      "loss": 0.6717,
      "step": 121
    },
    {
      "epoch": 0.77,
      "learning_rate": 1.743602633029255e-05,
      "loss": 0.6672,
      "step": 122
    },
    {
      "epoch": 0.78,
      "learning_rate": 1.7390089172206594e-05,
      "loss": 0.6634,
      "step": 123
    },
    {
      "epoch": 0.78,
      "learning_rate": 1.7343805817871885e-05,
      "loss": 0.6393,
      "step": 124
    },
    {
      "epoch": 0.79,
      "learning_rate": 1.729717843547927e-05,
      "loss": 0.6719,
      "step": 125
    },
    {
      "epoch": 0.79,
      "learning_rate": 1.725020920933593e-05,
      "loss": 0.7163,
      "step": 126
    },
    {
      "epoch": 0.8,
      "learning_rate": 1.7202900339763066e-05,
      "loss": 0.6749,
      "step": 127
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.7155254042992827e-05,
      "loss": 0.6229,
      "step": 128
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.710727255106447e-05,
      "loss": 0.6612,
      "step": 129
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.7058958111719836e-05,
      "loss": 0.6606,
      "step": 130
    },
    {
      "epoch": 0.83,
      "learning_rate": 1.7010312988297993e-05,
      "loss": 0.7066,
      "step": 131
    },
    {
      "epoch": 0.83,
      "learning_rate": 1.696133945962927e-05,
      "loss": 0.6702,
      "step": 132
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.691203981992845e-05,
      "loss": 0.6993,
      "step": 133
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.686241637868734e-05,
      "loss": 0.6994,
      "step": 134
    },
    {
      "epoch": 0.85,
      "learning_rate": 1.681247146056654e-05,
      "loss": 0.6377,
      "step": 135
    },
    {
      "epoch": 0.86,
      "learning_rate": 1.676220740528659e-05,
      "loss": 0.6326,
      "step": 136
    },
    {
      "epoch": 0.86,
      "learning_rate": 1.67116265675183e-05,
      "loss": 0.6737,
      "step": 137
    },
    {
      "epoch": 0.87,
      "learning_rate": 1.6660731316772503e-05,
      "loss": 0.6627,
      "step": 138
    },
    {
      "epoch": 0.88,
      "learning_rate": 1.660952403728902e-05,
      "loss": 0.6918,
      "step": 139
    },
    {
      "epoch": 0.88,
      "learning_rate": 1.655800712792498e-05,
      "loss": 0.7077,
      "step": 140
    },
    {
      "epoch": 0.89,
      "learning_rate": 1.650618300204242e-05,
      "loss": 0.6593,
      "step": 141
    },
    {
      "epoch": 0.9,
      "learning_rate": 1.6454054087395284e-05,
      "loss": 0.6868,
      "step": 142
    },
    {
      "epoch": 0.9,
      "learning_rate": 1.6401622826015616e-05,
      "loss": 0.6612,
      "step": 143
    },
    {
      "epoch": 0.91,
      "learning_rate": 1.634889167409923e-05,
      "loss": 0.6269,
      "step": 144
    },
    {
      "epoch": 0.91,
      "learning_rate": 1.6295863101890603e-05,
      "loss": 0.6932,
      "step": 145
    },
    {
      "epoch": 0.92,
      "learning_rate": 1.624253959356717e-05,
      "loss": 0.6717,
      "step": 146
    },
    {
      "epoch": 0.93,
      "learning_rate": 1.6188923647122946e-05,
      "loss": 0.7394,
      "step": 147
    },
    {
      "epoch": 0.93,
      "learning_rate": 1.613501777425152e-05,
      "loss": 0.61,
      "step": 148
    },
    {
      "epoch": 0.94,
      "learning_rate": 1.6080824500228367e-05,
      "loss": 0.7195,
      "step": 149
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.6026346363792565e-05,
      "loss": 0.6338,
      "step": 150
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.5971585917027864e-05,
      "loss": 0.7215,
      "step": 151
    },
    {
      "epoch": 0.96,
      "learning_rate": 1.5916545725243124e-05,
      "loss": 0.7139,
      "step": 152
    },
    {
      "epoch": 0.96,
      "learning_rate": 1.5861228366852148e-05,
      "loss": 0.6463,
      "step": 153
    },
    {
      "epoch": 0.97,
      "learning_rate": 1.5805636433252892e-05,
      "loss": 0.6277,
      "step": 154
    },
    {
      "epoch": 0.98,
      "learning_rate": 1.574977252870607e-05,
      "loss": 0.6515,
      "step": 155
    },
    {
      "epoch": 0.98,
      "learning_rate": 1.5693639270213138e-05,
      "loss": 0.6448,
      "step": 156
    },
    {
      "epoch": 0.99,
      "learning_rate": 1.5637239287393725e-05,
      "loss": 0.675,
      "step": 157
    },
    {
      "epoch": 1.0,
      "learning_rate": 1.5580575222362435e-05,
      "loss": 0.6509,
      "step": 158
    },
    {
      "epoch": 1.0,
      "learning_rate": 1.552364972960506e-05,
      "loss": 0.632,
      "step": 159
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.5466465475854246e-05,
      "loss": 0.4572,
      "step": 160
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.540902513996456e-05,
      "loss": 0.435,
      "step": 161
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.5351331412787004e-05,
      "loss": 0.4404,
      "step": 162
    },
    {
      "epoch": 1.03,
      "learning_rate": 1.5293386997042943e-05,
      "loss": 0.4234,
      "step": 163
    },
    {
      "epoch": 1.03,
      "learning_rate": 1.5235194607197508e-05,
      "loss": 0.456,
      "step": 164
    },
    {
      "epoch": 1.04,
      "learning_rate": 1.5176756969332428e-05,
      "loss": 0.4146,
      "step": 165
    },
    {
      "epoch": 1.05,
      "learning_rate": 1.5118076821018322e-05,
      "loss": 0.4506,
      "step": 166
    },
    {
      "epoch": 1.05,
      "learning_rate": 1.5059156911186465e-05,
      "loss": 0.4413,
      "step": 167
    },
    {
      "epoch": 1.06,
      "learning_rate": 1.5000000000000002e-05,
      "loss": 0.4192,
      "step": 168
    },
    {
      "epoch": 1.07,
      "learning_rate": 1.494060885872464e-05,
      "loss": 0.4308,
      "step": 169
    },
    {
      "epoch": 1.07,
      "learning_rate": 1.488098626959885e-05,
      "loss": 0.4128,
      "step": 170
    },
    {
      "epoch": 1.08,
      "learning_rate": 1.4821135025703491e-05,
      "loss": 0.4333,
      "step": 171
    },
    {
      "epoch": 1.08,
      "learning_rate": 1.4761057930831002e-05,
      "loss": 0.4335,
      "step": 172
    },
    {
      "epoch": 1.09,
      "learning_rate": 1.470075779935404e-05,
      "loss": 0.4216,
      "step": 173
    },
    {
      "epoch": 1.1,
      "learning_rate": 1.4640237456093636e-05,
      "loss": 0.4057,
      "step": 174
    },
    {
      "epoch": 1.1,
      "learning_rate": 1.4579499736186864e-05,
      "loss": 0.4344,
      "step": 175
    },
    {
      "epoch": 1.11,
      "learning_rate": 1.4518547484954033e-05,
      "loss": 0.4579,
      "step": 176
    },
    {
      "epoch": 1.12,
      "learning_rate": 1.4457383557765385e-05,
      "loss": 0.4115,
      "step": 177
    },
    {
      "epoch": 1.12,
      "learning_rate": 1.439601081990734e-05,
      "loss": 0.4154,
      "step": 178
    },
    {
      "epoch": 1.13,
      "learning_rate": 1.4334432146448272e-05,
      "loss": 0.402,
      "step": 179
    },
    {
      "epoch": 1.13,
      "learning_rate": 1.427265042210381e-05,
      "loss": 0.3958,
      "step": 180
    },
    {
      "epoch": 1.14,
      "learning_rate": 1.4210668541101713e-05,
      "loss": 0.4205,
      "step": 181
    },
    {
      "epoch": 1.15,
      "learning_rate": 1.4148489407046274e-05,
      "loss": 0.4091,
      "step": 182
    },
    {
      "epoch": 1.15,
      "learning_rate": 1.4086115932782316e-05,
      "loss": 0.3839,
      "step": 183
    },
    {
      "epoch": 1.16,
      "learning_rate": 1.4023551040258726e-05,
      "loss": 0.4377,
      "step": 184
    },
    {
      "epoch": 1.17,
      "learning_rate": 1.396079766039157e-05,
      "loss": 0.4185,
      "step": 185
    },
    {
      "epoch": 1.17,
      "learning_rate": 1.3897858732926794e-05,
      "loss": 0.4318,
      "step": 186
    },
    {
      "epoch": 1.18,
      "learning_rate": 1.3834737206302519e-05,
      "loss": 0.4272,
      "step": 187
    },
    {
      "epoch": 1.18,
      "learning_rate": 1.3771436037510897e-05,
      "loss": 0.4168,
      "step": 188
    },
    {
      "epoch": 1.19,
      "learning_rate": 1.3707958191959609e-05,
      "loss": 0.4151,
      "step": 189
    },
    {
      "epoch": 1.2,
      "learning_rate": 1.3644306643332939e-05,
      "loss": 0.4008,
      "step": 190
    },
    {
      "epoch": 1.2,
      "learning_rate": 1.3580484373452462e-05,
      "loss": 0.4007,
      "step": 191
    },
    {
      "epoch": 1.21,
      "learning_rate": 1.3516494372137368e-05,
      "loss": 0.397,
      "step": 192
    },
    {
      "epoch": 1.22,
      "learning_rate": 1.34523396370644e-05,
      "loss": 0.4091,
      "step": 193
    },
    {
      "epoch": 1.22,
      "learning_rate": 1.3388023173627413e-05,
      "loss": 0.4383,
      "step": 194
    },
    {
      "epoch": 1.23,
      "learning_rate": 1.3323547994796597e-05,
      "loss": 0.4,
      "step": 195
    },
    {
      "epoch": 1.24,
      "learning_rate": 1.3258917120977327e-05,
      "loss": 0.4082,
      "step": 196
    },
    {
      "epoch": 1.24,
      "learning_rate": 1.3194133579868672e-05,
      "loss": 0.4201,
      "step": 197
    },
    {
      "epoch": 1.25,
      "learning_rate": 1.3129200406321545e-05,
      "loss": 0.4263,
      "step": 198
    },
    {
      "epoch": 1.25,
      "learning_rate": 1.3064120642196549e-05,
      "loss": 0.4405,
      "step": 199
    },
    {
      "epoch": 1.26,
      "learning_rate": 1.299889733622147e-05,
      "loss": 0.4453,
      "step": 200
    },
    {
      "epoch": 1.27,
      "learning_rate": 1.2933533543848462e-05,
      "loss": 0.4142,
      "step": 201
    },
    {
      "epoch": 1.27,
      "learning_rate": 1.2868032327110904e-05,
      "loss": 0.4161,
      "step": 202
    },
    {
      "epoch": 1.28,
      "learning_rate": 1.2802396754479958e-05,
      "loss": 0.4122,
      "step": 203
    },
    {
      "epoch": 1.29,
      "learning_rate": 1.2736629900720832e-05,
      "loss": 0.3909,
      "step": 204
    },
    {
      "epoch": 1.29,
      "learning_rate": 1.2670734846748717e-05,
      "loss": 0.3958,
      "step": 205
    },
    {
      "epoch": 1.3,
      "learning_rate": 1.260471467948449e-05,
      "loss": 0.425,
      "step": 206
    },
    {
      "epoch": 1.3,
      "learning_rate": 1.2538572491710079e-05,
      "loss": 0.4134,
      "step": 207
    },
    {
      "epoch": 1.31,
      "learning_rate": 1.247231138192359e-05,
      "loss": 0.427,
      "step": 208
    },
    {
      "epoch": 1.32,
      "learning_rate": 1.2405934454194146e-05,
      "loss": 0.4055,
      "step": 209
    },
    {
      "epoch": 1.32,
      "learning_rate": 1.2339444818016488e-05,
      "loss": 0.4106,
      "step": 210
    },
    {
      "epoch": 1.33,
      "learning_rate": 1.227284558816529e-05,
      "loss": 0.4315,
      "step": 211
    },
    {
      "epoch": 1.34,
      "learning_rate": 1.220613988454926e-05,
      "loss": 0.4516,
      "step": 212
    },
    {
      "epoch": 1.34,
      "learning_rate": 1.2139330832064975e-05,
      "loss": 0.4051,
      "step": 213
    },
    {
      "epoch": 1.35,
      "learning_rate": 1.2072421560450497e-05,
      "loss": 0.4188,
      "step": 214
    },
    {
      "epoch": 1.36,
      "learning_rate": 1.2005415204138753e-05,
      "loss": 0.4053,
      "step": 215
    },
    {
      "epoch": 1.36,
      "learning_rate": 1.1938314902110701e-05,
      "loss": 0.4075,
      "step": 216
    },
    {
      "epoch": 1.37,
      "learning_rate": 1.1871123797748285e-05,
      "loss": 0.3825,
      "step": 217
    },
    {
      "epoch": 1.37,
      "learning_rate": 1.1803845038687171e-05,
      "loss": 0.3963,
      "step": 218
    },
    {
      "epoch": 1.38,
      "learning_rate": 1.1736481776669307e-05,
      "loss": 0.3849,
      "step": 219
    },
    {
      "epoch": 1.39,
      "learning_rate": 1.1669037167395256e-05,
      "loss": 0.4035,
      "step": 220
    },
    {
      "epoch": 1.39,
      "learning_rate": 1.160151437037639e-05,
      "loss": 0.4004,
      "step": 221
    },
    {
      "epoch": 1.4,
      "learning_rate": 1.1533916548786856e-05,
      "loss": 0.4409,
      "step": 222
    },
    {
      "epoch": 1.41,
      "learning_rate": 1.1466246869315407e-05,
      "loss": 0.3907,
      "step": 223
    },
    {
      "epoch": 1.41,
      "learning_rate": 1.1398508502017047e-05,
      "loss": 0.4235,
      "step": 224
    },
    {
      "epoch": 1.42,
      "learning_rate": 1.133070462016454e-05,
      "loss": 0.4046,
      "step": 225
    },
    {
      "epoch": 1.42,
      "learning_rate": 1.1262838400099733e-05,
      "loss": 0.4355,
      "step": 226
    },
    {
      "epoch": 1.43,
      "learning_rate": 1.119491302108479e-05,
      "loss": 0.3875,
      "step": 227
    },
    {
      "epoch": 1.44,
      "learning_rate": 1.1126931665153213e-05,
      "loss": 0.4157,
      "step": 228
    },
    {
      "epoch": 1.44,
      "learning_rate": 1.1058897516960817e-05,
      "loss": 0.4173,
      "step": 229
    },
    {
      "epoch": 1.45,
      "learning_rate": 1.0990813763636511e-05,
      "loss": 0.4087,
      "step": 230
    },
    {
      "epoch": 1.46,
      "learning_rate": 1.092268359463302e-05,
      "loss": 0.4305,
      "step": 231
    },
    {
      "epoch": 1.46,
      "learning_rate": 1.0854510201577451e-05,
      "loss": 0.4115,
      "step": 232
    },
    {
      "epoch": 1.47,
      "learning_rate": 1.0786296778121787e-05,
      "loss": 0.4193,
      "step": 233
    },
    {
      "epoch": 1.47,
      "learning_rate": 1.0718046519793276e-05,
      "loss": 0.4337,
      "step": 234
    },
    {
      "epoch": 1.48,
      "learning_rate": 1.0649762623844733e-05,
      "loss": 0.4085,
      "step": 235
    },
    {
      "epoch": 1.49,
      "learning_rate": 1.0581448289104759e-05,
      "loss": 0.3991,
      "step": 236
    },
    {
      "epoch": 1.49,
      "learning_rate": 1.0513106715827897e-05,
      "loss": 0.419,
      "step": 237
    },
    {
      "epoch": 1.5,
      "learning_rate": 1.0444741105544705e-05,
      "loss": 0.4305,
      "step": 238
    },
    {
      "epoch": 1.51,
      "learning_rate": 1.0376354660911772e-05,
      "loss": 0.4014,
      "step": 239
    },
    {
      "epoch": 1.51,
      "learning_rate": 1.0307950585561705e-05,
      "loss": 0.4348,
      "step": 240
    },
    {
      "epoch": 1.52,
      "learning_rate": 1.0239532083953032e-05,
      "loss": 0.3982,
      "step": 241
    },
    {
      "epoch": 1.53,
      "learning_rate": 1.0171102361220093e-05,
      "loss": 0.4035,
      "step": 242
    },
    {
      "epoch": 1.53,
      "learning_rate": 1.01026646230229e-05,
      "loss": 0.4073,
      "step": 243
    },
    {
      "epoch": 1.54,
      "learning_rate": 1.0034222075396954e-05,
      "loss": 0.401,
      "step": 244
    },
    {
      "epoch": 1.54,
      "learning_rate": 9.965777924603053e-06,
      "loss": 0.4082,
      "step": 245
    },
    {
      "epoch": 1.55,
      "learning_rate": 9.897335376977104e-06,
      "loss": 0.4418,
      "step": 246
    },
    {
      "epoch": 1.56,
      "learning_rate": 9.828897638779909e-06,
      "loss": 0.3685,
      "step": 247
    },
    {
      "epoch": 1.56,
      "learning_rate": 9.760467916046971e-06,
      "loss": 0.4238,
      "step": 248
    },
    {
      "epoch": 1.57,
      "learning_rate": 9.692049414438298e-06,
      "loss": 0.4193,
      "step": 249
    },
    {
      "epoch": 1.58,
      "learning_rate": 9.62364533908823e-06,
      "loss": 0.4036,
      "step": 250
    },
    {
      "epoch": 1.58,
      "learning_rate": 9.555258894455298e-06,
      "loss": 0.4095,
      "step": 251
    },
    {
      "epoch": 1.59,
      "learning_rate": 9.486893284172103e-06,
      "loss": 0.4116,
      "step": 252
    },
    {
      "epoch": 1.59,
      "learning_rate": 9.418551710895243e-06,
      "loss": 0.4196,
      "step": 253
    },
    {
      "epoch": 1.6,
      "learning_rate": 9.350237376155269e-06,
      "loss": 0.4277,
      "step": 254
    },
    {
      "epoch": 1.61,
      "learning_rate": 9.281953480206725e-06,
      "loss": 0.4059,
      "step": 255
    },
    {
      "epoch": 1.61,
      "learning_rate": 9.213703221878217e-06,
      "loss": 0.4108,
      "step": 256
    },
    {
      "epoch": 1.62,
      "learning_rate": 9.14548979842255e-06,
      "loss": 0.4119,
      "step": 257
    },
    {
      "epoch": 1.63,
      "learning_rate": 9.07731640536698e-06,
      "loss": 0.418,
      "step": 258
    },
    {
      "epoch": 1.63,
      "learning_rate": 9.00918623636349e-06,
      "loss": 0.3981,
      "step": 259
    },
    {
      "epoch": 1.64,
      "learning_rate": 8.941102483039188e-06,
      "loss": 0.4068,
      "step": 260
    },
    {
      "epoch": 1.65,
      "learning_rate": 8.87306833484679e-06,
      "loss": 0.4149,
      "step": 261
    },
    {
      "epoch": 1.65,
      "learning_rate": 8.805086978915215e-06,
      "loss": 0.4364,
      "step": 262
    },
    {
      "epoch": 1.66,
      "learning_rate": 8.737161599900267e-06,
      "loss": 0.4082,
      "step": 263
    },
    {
      "epoch": 1.66,
      "learning_rate": 8.669295379835467e-06,
      "loss": 0.4084,
      "step": 264
    },
    {
      "epoch": 1.67,
      "learning_rate": 8.601491497982956e-06,
      "loss": 0.3942,
      "step": 265
    },
    {
      "epoch": 1.68,
      "learning_rate": 8.533753130684596e-06,
      "loss": 0.412,
      "step": 266
    },
    {
      "epoch": 1.68,
      "learning_rate": 8.466083451213145e-06,
      "loss": 0.4041,
      "step": 267
    },
    {
      "epoch": 1.69,
      "learning_rate": 8.398485629623613e-06,
      "loss": 0.4029,
      "step": 268
    },
    {
      "epoch": 1.7,
      "learning_rate": 8.330962832604747e-06,
      "loss": 0.4117,
      "step": 269
    },
    {
      "epoch": 1.7,
      "learning_rate": 8.263518223330698e-06,
      "loss": 0.3922,
      "step": 270
    },
    {
      "epoch": 1.71,
      "learning_rate": 8.19615496131283e-06,
      "loss": 0.4185,
      "step": 271
    },
    {
      "epoch": 1.71,
      "learning_rate": 8.128876202251719e-06,
      "loss": 0.4326,
      "step": 272
    },
    {
      "epoch": 1.72,
      "learning_rate": 8.0616850978893e-06,
      "loss": 0.4173,
      "step": 273
    },
    {
      "epoch": 1.73,
      "learning_rate": 7.994584795861248e-06,
      "loss": 0.3945,
      "step": 274
    },
    {
      "epoch": 1.73,
      "learning_rate": 7.927578439549506e-06,
      "loss": 0.4228,
      "step": 275
    },
    {
      "epoch": 1.74,
      "learning_rate": 7.860669167935028e-06,
      "loss": 0.4169,
      "step": 276
    },
    {
      "epoch": 1.75,
      "learning_rate": 7.793860115450744e-06,
      "loss": 0.406,
      "step": 277
    },
    {
      "epoch": 1.75,
      "learning_rate": 7.727154411834712e-06,
      "loss": 0.4343,
      "step": 278
    },
    {
      "epoch": 1.76,
      "learning_rate": 7.660555181983517e-06,
      "loss": 0.4037,
      "step": 279
    },
    {
      "epoch": 1.76,
      "learning_rate": 7.5940655458058575e-06,
      "loss": 0.388,
      "step": 280
    },
    {
      "epoch": 1.77,
      "learning_rate": 7.527688618076413e-06,
      "loss": 0.4398,
      "step": 281
    },
    {
      "epoch": 1.78,
      "learning_rate": 7.461427508289922e-06,
      "loss": 0.3886,
      "step": 282
    },
    {
      "epoch": 1.78,
      "learning_rate": 7.395285320515513e-06,
      "loss": 0.4159,
      "step": 283
    },
    {
      "epoch": 1.79,
      "learning_rate": 7.329265153251285e-06,
      "loss": 0.429,
      "step": 284
    },
    {
      "epoch": 1.8,
      "learning_rate": 7.263370099279173e-06,
      "loss": 0.4257,
      "step": 285
    },
    {
      "epoch": 1.8,
      "learning_rate": 7.197603245520042e-06,
      "loss": 0.4306,
      "step": 286
    },
    {
      "epoch": 1.81,
      "learning_rate": 7.131967672889101e-06,
      "loss": 0.4098,
      "step": 287
    },
    {
      "epoch": 1.82,
      "learning_rate": 7.066466456151541e-06,
      "loss": 0.3874,
      "step": 288
    },
    {
      "epoch": 1.82,
      "learning_rate": 7.001102663778533e-06,
      "loss": 0.4355,
      "step": 289
    },
    {
      "epoch": 1.83,
      "learning_rate": 6.935879357803453e-06,
      "loss": 0.416,
      "step": 290
    },
    {
      "epoch": 1.83,
      "learning_rate": 6.870799593678459e-06,
      "loss": 0.4404,
      "step": 291
    },
    {
      "epoch": 1.84,
      "learning_rate": 6.80586642013133e-06,
      "loss": 0.4473,
      "step": 292
    },
    {
      "epoch": 1.85,
      "learning_rate": 6.741082879022671e-06,
      "loss": 0.4038,
      "step": 293
    },
    {
      "epoch": 1.85,
      "learning_rate": 6.6764520052034054e-06,
      "loss": 0.4069,
      "step": 294
    },
    {
      "epoch": 1.86,
      "learning_rate": 6.61197682637259e-06,
      "loss": 0.4313,
      "step": 295
    },
    {
      "epoch": 1.87,
      "learning_rate": 6.547660362935603e-06,
      "loss": 0.4478,
      "step": 296
    },
    {
      "epoch": 1.87,
      "learning_rate": 6.483505627862632e-06,
      "loss": 0.424,
      "step": 297
    },
    {
      "epoch": 1.88,
      "learning_rate": 6.419515626547543e-06,
      "loss": 0.4191,
      "step": 298
    },
    {
      "epoch": 1.88,
      "learning_rate": 6.3556933566670656e-06,
      "loss": 0.4187,
      "step": 299
    },
    {
      "epoch": 1.89,
      "learning_rate": 6.292041808040393e-06,
      "loss": 0.4071,
      "step": 300
    },
    {
      "epoch": 1.9,
      "learning_rate": 6.228563962489106e-06,
      "loss": 0.409,
      "step": 301
    },
    {
      "epoch": 1.9,
      "learning_rate": 6.165262793697486e-06,
      "loss": 0.4286,
      "step": 302
    },
    {
      "epoch": 1.91,
      "learning_rate": 6.102141267073207e-06,
      "loss": 0.413,
      "step": 303
    },
    {
      "epoch": 1.92,
      "learning_rate": 6.039202339608432e-06,
      "loss": 0.4077,
      "step": 304
    },
    {
      "epoch": 1.92,
      "learning_rate": 5.9764489597412744e-06,
      "loss": 0.4276,
      "step": 305
    },
    {
      "epoch": 1.93,
      "learning_rate": 5.913884067217686e-06,
      "loss": 0.4346,
      "step": 306
    },
    {
      "epoch": 1.94,
      "learning_rate": 5.851510592953729e-06,
      "loss": 0.4156,
      "step": 307
    },
    {
      "epoch": 1.94,
      "learning_rate": 5.7893314588982905e-06,
      "loss": 0.4117,
      "step": 308
    },
    {
      "epoch": 1.95,
      "learning_rate": 5.727349577896194e-06,
      "loss": 0.4142,
      "step": 309
    },
    {
      "epoch": 1.95,
      "learning_rate": 5.66556785355173e-06,
      "loss": 0.4249,
      "step": 310
    },
    {
      "epoch": 1.96,
      "learning_rate": 5.603989180092661e-06,
      "loss": 0.4039,
      "step": 311
    },
    {
      "epoch": 1.97,
      "learning_rate": 5.542616442234618e-06,
      "loss": 0.4243,
      "step": 312
    },
    {
      "epoch": 1.97,
      "learning_rate": 5.481452515045974e-06,
      "loss": 0.4111,
      "step": 313
    },
    {
      "epoch": 1.98,
      "learning_rate": 5.420500263813141e-06,
      "loss": 0.4139,
      "step": 314
    },
    {
      "epoch": 1.99,
      "learning_rate": 5.3597625439063685e-06,
      "loss": 0.4096,
      "step": 315
    },
    {
      "epoch": 1.99,
      "learning_rate": 5.299242200645959e-06,
      "loss": 0.3924,
      "step": 316
    },
    {
      "epoch": 2.0,
      "learning_rate": 5.238942069169e-06,
      "loss": 0.4286,
      "step": 317
    },
    {
      "epoch": 2.0,
      "learning_rate": 5.178864974296511e-06,
      "loss": 0.3065,
      "step": 318
    },
    {
      "epoch": 2.01,
      "learning_rate": 5.119013730401152e-06,
      "loss": 0.2903,
      "step": 319
    },
    {
      "epoch": 2.02,
      "learning_rate": 5.059391141275358e-06,
      "loss": 0.2945,
      "step": 320
    },
    {
      "epoch": 2.02,
      "learning_rate": 5.000000000000003e-06,
      "loss": 0.2773,
      "step": 321
    },
    {
      "epoch": 2.03,
      "learning_rate": 4.940843088813537e-06,
      "loss": 0.2602,
      "step": 322
    },
    {
      "epoch": 2.04,
      "learning_rate": 4.881923178981681e-06,
      "loss": 0.2753,
      "step": 323
    },
    {
      "epoch": 2.04,
      "learning_rate": 4.823243030667576e-06,
      "loss": 0.2545,
      "step": 324
    },
    {
      "epoch": 2.05,
      "learning_rate": 4.764805392802497e-06,
      "loss": 0.2758,
      "step": 325
    },
    {
      "epoch": 2.05,
      "learning_rate": 4.70661300295706e-06,
      "loss": 0.2496,
      "step": 326
    },
    {
      "epoch": 2.06,
      "learning_rate": 4.648668587212998e-06,
      "loss": 0.2853,
      "step": 327
    },
    {
      "epoch": 2.07,
      "learning_rate": 4.5909748600354395e-06,
      "loss": 0.2748,
      "step": 328
    },
    {
      "epoch": 2.07,
      "learning_rate": 4.533534524145756e-06,
      "loss": 0.2689,
      "step": 329
    },
    {
      "epoch": 2.08,
      "learning_rate": 4.476350270394942e-06,
      "loss": 0.2692,
      "step": 330
    },
    {
      "epoch": 2.09,
      "learning_rate": 4.419424777637565e-06,
      "loss": 0.2672,
      "step": 331
    },
    {
      "epoch": 2.09,
      "learning_rate": 4.362760712606278e-06,
      "loss": 0.2747,
      "step": 332
    },
    {
      "epoch": 2.1,
      "learning_rate": 4.306360729786867e-06,
      "loss": 0.2728,
      "step": 333
    },
    {
      "epoch": 2.11,
      "learning_rate": 4.2502274712939355e-06,
      "loss": 0.2904,
      "step": 334
    },
    {
      "epoch": 2.11,
      "learning_rate": 4.1943635667471095e-06,
      "loss": 0.2502,
      "step": 335
    },
    {
      "epoch": 2.12,
      "learning_rate": 4.138771633147856e-06,
      "loss": 0.295,
      "step": 336
    },
    {
      "epoch": 2.12,
      "learning_rate": 4.083454274756881e-06,
      "loss": 0.2526,
      "step": 337
    },
    {
      "epoch": 2.13,
      "learning_rate": 4.028414082972141e-06,
      "loss": 0.2534,
      "step": 338
    },
    {
      "epoch": 2.14,
      "learning_rate": 3.973653636207437e-06,
      "loss": 0.2668,
      "step": 339
    },
    {
      "epoch": 2.14,
      "learning_rate": 3.919175499771635e-06,
      "loss": 0.2513,
      "step": 340
    },
    {
      "epoch": 2.15,
      "learning_rate": 3.864982225748481e-06,
      "loss": 0.2498,
      "step": 341
    },
    {
      "epoch": 2.16,
      "learning_rate": 3.8110763528770543e-06,
      "loss": 0.2415,
      "step": 342
    },
    {
      "epoch": 2.16,
      "learning_rate": 3.7574604064328336e-06,
      "loss": 0.2417,
      "step": 343
    },
    {
      "epoch": 2.17,
      "learning_rate": 3.704136898109403e-06,
      "loss": 0.2625,
      "step": 344
    },
    {
      "epoch": 2.17,
      "learning_rate": 3.651108325900773e-06,
      "loss": 0.2494,
      "step": 345
    },
    {
      "epoch": 2.18,
      "learning_rate": 3.5983771739843855e-06,
      "loss": 0.2542,
      "step": 346
    },
    {
      "epoch": 2.19,
      "learning_rate": 3.5459459126047226e-06,
      "loss": 0.2594,
      "step": 347
    },
    {
      "epoch": 2.19,
      "learning_rate": 3.493816997957582e-06,
      "loss": 0.257,
      "step": 348
    },
    {
      "epoch": 2.2,
      "learning_rate": 3.4419928720750274e-06,
      "loss": 0.2541,
      "step": 349
    },
    {
      "epoch": 2.21,
      "learning_rate": 3.3904759627109828e-06,
      "loss": 0.2547,
      "step": 350
    },
    {
      "epoch": 2.21,
      "learning_rate": 3.339268683227499e-06,
      "loss": 0.2385,
      "step": 351
    },
    {
      "epoch": 2.22,
      "learning_rate": 3.288373432481703e-06,
      "loss": 0.2701,
      "step": 352
    },
    {
      "epoch": 2.22,
      "learning_rate": 3.2377925947134137e-06,
      "loss": 0.2534,
      "step": 353
    },
    {
      "epoch": 2.23,
      "learning_rate": 3.1875285394334575e-06,
      "loss": 0.2472,
      "step": 354
    },
    {
      "epoch": 2.24,
      "learning_rate": 3.1375836213126653e-06,
      "loss": 0.2639,
      "step": 355
    },
    {
      "epoch": 2.24,
      "learning_rate": 3.087960180071553e-06,
      "loss": 0.2749,
      "step": 356
    },
    {
      "epoch": 2.25,
      "learning_rate": 3.0386605403707347e-06,
      "loss": 0.2694,
      "step": 357
    },
    {
      "epoch": 2.26,
      "learning_rate": 2.9896870117020073e-06,
      "loss": 0.2471,
      "step": 358
    },
    {
      "epoch": 2.26,
      "learning_rate": 2.9410418882801682e-06,
      "loss": 0.2799,
      "step": 359
    },
    {
      "epoch": 2.27,
      "learning_rate": 2.8927274489355296e-06,
      "loss": 0.2504,
      "step": 360
    },
    {
      "epoch": 2.28,
      "learning_rate": 2.844745957007178e-06,
      "loss": 0.2726,
      "step": 361
    },
    {
      "epoch": 2.28,
      "learning_rate": 2.797099660236937e-06,
      "loss": 0.2854,
      "step": 362
    },
    {
      "epoch": 2.29,
      "learning_rate": 2.749790790664074e-06,
      "loss": 0.259,
      "step": 363
    },
    {
      "epoch": 2.29,
      "learning_rate": 2.702821564520732e-06,
      "loss": 0.2693,
      "step": 364
    },
    {
      "epoch": 2.3,
      "learning_rate": 2.6561941821281145e-06,
      "loss": 0.2449,
      "step": 365
    },
    {
      "epoch": 2.31,
      "learning_rate": 2.6099108277934105e-06,
      "loss": 0.2593,
      "step": 366
    },
    {
      "epoch": 2.31,
      "learning_rate": 2.5639736697074525e-06,
      "loss": 0.2709,
      "step": 367
    },
    {
      "epoch": 2.32,
      "learning_rate": 2.518384859843168e-06,
      "loss": 0.2645,
      "step": 368
    },
    {
      "epoch": 2.33,
      "learning_rate": 2.4731465338547556e-06,
      "loss": 0.254,
      "step": 369
    },
    {
      "epoch": 2.33,
      "learning_rate": 2.428260810977641e-06,
      "loss": 0.2508,
      "step": 370
    },
    {
      "epoch": 2.34,
      "learning_rate": 2.3837297939291893e-06,
      "loss": 0.2675,
      "step": 371
    },
    {
      "epoch": 2.34,
      "learning_rate": 2.339555568810221e-06,
      "loss": 0.2585,
      "step": 372
    },
    {
      "epoch": 2.35,
      "learning_rate": 2.2957402050072717e-06,
      "loss": 0.3071,
      "step": 373
    },
    {
      "epoch": 2.36,
      "learning_rate": 2.252285755095652e-06,
      "loss": 0.2604,
      "step": 374
    },
    {
      "epoch": 2.36,
      "learning_rate": 2.209194254743295e-06,
      "loss": 0.2635,
      "step": 375
    },
    {
      "epoch": 2.37,
      "learning_rate": 2.166467722615394e-06,
      "loss": 0.2562,
      "step": 376
    },
    {
      "epoch": 2.38,
      "learning_rate": 2.124108160279832e-06,
      "loss": 0.2638,
      "step": 377
    },
    {
      "epoch": 2.38,
      "learning_rate": 2.0821175521134208e-06,
      "loss": 0.2689,
      "step": 378
    },
    {
      "epoch": 2.39,
      "learning_rate": 2.0404978652089325e-06,
      "loss": 0.253,
      "step": 379
    },
    {
      "epoch": 2.4,
      "learning_rate": 1.999251049282962e-06,
      "loss": 0.2746,
      "step": 380
    },
    {
      "epoch": 2.4,
      "learning_rate": 1.9583790365845823e-06,
      "loss": 0.2737,
      "step": 381
    },
    {
      "epoch": 2.41,
      "learning_rate": 1.917883741804829e-06,
      "loss": 0.2533,
      "step": 382
    },
    {
      "epoch": 2.41,
      "learning_rate": 1.877767061986997e-06,
      "loss": 0.2461,
      "step": 383
    },
    {
      "epoch": 2.42,
      "learning_rate": 1.8380308764377841e-06,
      "loss": 0.2637,
      "step": 384
    },
    {
      "epoch": 2.43,
      "learning_rate": 1.7986770466392445e-06,
      "loss": 0.273,
      "step": 385
    },
    {
      "epoch": 2.43,
      "learning_rate": 1.75970741616159e-06,
      "loss": 0.2691,
      "step": 386
    },
    {
      "epoch": 2.44,
      "learning_rate": 1.7211238105768213e-06,
      "loss": 0.2659,
      "step": 387
    },
    {
      "epoch": 2.45,
      "learning_rate": 1.6829280373732126e-06,
      "loss": 0.263,
      "step": 388
    },
    {
      "epoch": 2.45,
      "learning_rate": 1.6451218858706374e-06,
      "loss": 0.2622,
      "step": 389
    },
    {
      "epoch": 2.46,
      "learning_rate": 1.607707127136734e-06,
      "loss": 0.2524,
      "step": 390
    },
    {
      "epoch": 2.46,
      "learning_rate": 1.57068551390396e-06,
      "loss": 0.2765,
      "step": 391
    },
    {
      "epoch": 2.47,
      "learning_rate": 1.5340587804874662e-06,
      "loss": 0.2485,
      "step": 392
    },
    {
      "epoch": 2.48,
      "learning_rate": 1.4978286427038602e-06,
      "loss": 0.2543,
      "step": 393
    },
    {
      "epoch": 2.48,
      "learning_rate": 1.4619967977908157e-06,
      "loss": 0.2524,
      "step": 394
    },
    {
      "epoch": 2.49,
      "learning_rate": 1.4265649243275782e-06,
      "loss": 0.274,
      "step": 395
    },
    {
      "epoch": 2.5,
      "learning_rate": 1.3915346821563235e-06,
      "loss": 0.2585,
      "step": 396
    },
    {
      "epoch": 2.5,
      "learning_rate": 1.3569077123043973e-06,
      "loss": 0.2535,
      "step": 397
    },
    {
      "epoch": 2.51,
      "learning_rate": 1.322685636907447e-06,
      "loss": 0.2546,
      "step": 398
    },
    {
      "epoch": 2.51,
      "learning_rate": 1.2888700591334225e-06,
      "loss": 0.2408,
      "step": 399
    },
    {
      "epoch": 2.52,
      "learning_rate": 1.2554625631074846e-06,
      "loss": 0.256,
      "step": 400
    },
    {
      "epoch": 2.53,
      "learning_rate": 1.2224647138377854e-06,
      "loss": 0.2545,
      "step": 401
    },
    {
      "epoch": 2.53,
      "learning_rate": 1.1898780571421554e-06,
      "loss": 0.2623,
      "step": 402
    },
    {
      "epoch": 2.54,
      "learning_rate": 1.1577041195756954e-06,
      "loss": 0.2696,
      "step": 403
    },
    {
      "epoch": 2.55,
      "learning_rate": 1.1259444083592585e-06,
      "loss": 0.2458,
      "step": 404
    },
    {
      "epoch": 2.55,
      "learning_rate": 1.0946004113088381e-06,
      "loss": 0.2816,
      "step": 405
    },
    {
      "epoch": 2.56,
      "learning_rate": 1.0636735967658785e-06,
      "loss": 0.2945,
      "step": 406
    },
    {
      "epoch": 2.57,
      "learning_rate": 1.033165413528483e-06,
      "loss": 0.246,
      "step": 407
    },
    {
      "epoch": 2.57,
      "learning_rate": 1.0030772907835484e-06,
      "loss": 0.2726,
      "step": 408
    },
    {
      "epoch": 2.58,
      "learning_rate": 9.734106380398022e-07,
      "loss": 0.252,
      "step": 409
    },
    {
      "epoch": 2.58,
      "learning_rate": 9.441668450617924e-07,
      "loss": 0.2643,
      "step": 410
    },
    {
      "epoch": 2.59,
      "learning_rate": 9.153472818047627e-07,
      "loss": 0.2737,
      "step": 411
    },
    {
      "epoch": 2.6,
      "learning_rate": 8.869532983504859e-07,
      "loss": 0.2644,
      "step": 412
    },
    {
      "epoch": 2.6,
      "learning_rate": 8.58986224844014e-07,
      "loss": 0.2461,
      "step": 413
    },
    {
      "epoch": 2.61,
      "learning_rate": 8.31447371431372e-07,
      "loss": 0.2749,
      "step": 414
    },
    {
      "epoch": 2.62,
      "learning_rate": 8.043380281981739e-07,
      "loss": 0.258,
      "step": 415
    },
    {
      "epoch": 2.62,
      "learning_rate": 7.776594651091995e-07,
      "loss": 0.2427,
      "step": 416
    },
    {
      "epoch": 2.63,
      "learning_rate": 7.514129319488839e-07,
      "loss": 0.2555,
      "step": 417
    },
    {
      "epoch": 2.63,
      "learning_rate": 7.255996582627878e-07,
      "loss": 0.2631,
      "step": 418
    },
    {
      "epoch": 2.64,
      "learning_rate": 7.002208532999933e-07,
      "loss": 0.2846,
      "step": 419
    },
    {
      "epoch": 2.65,
      "learning_rate": 6.752777059564431e-07,
      "loss": 0.272,
      "step": 420
    },
    {
      "epoch": 2.65,
      "learning_rate": 6.507713847192643e-07,
      "loss": 0.2343,
      "step": 421
    },
    {
      "epoch": 2.66,
      "learning_rate": 6.267030376120154e-07,
      "loss": 0.249,
      "step": 422
    },
    {
      "epoch": 2.67,
      "learning_rate": 6.030737921409169e-07,
      "loss": 0.2658,
      "step": 423
    },
    {
      "epoch": 2.67,
      "learning_rate": 5.798847552420184e-07,
      "loss": 0.2564,
      "step": 424
    },
    {
      "epoch": 2.68,
      "learning_rate": 5.571370132293552e-07,
      "loss": 0.2397,
      "step": 425
    },
    {
      "epoch": 2.69,
      "learning_rate": 5.348316317440549e-07,
      "loss": 0.263,
      "step": 426
    },
    {
      "epoch": 2.69,
      "learning_rate": 5.129696557044173e-07,
      "loss": 0.2594,
      "step": 427
    },
    {
      "epoch": 2.7,
      "learning_rate": 4.915521092569553e-07,
      "loss": 0.2341,
      "step": 428
    },
    {
      "epoch": 2.7,
      "learning_rate": 4.7057999572843516e-07,
      "loss": 0.2755,
      "step": 429
    },
    {
      "epoch": 2.71,
      "learning_rate": 4.50054297578858e-07,
      "loss": 0.2541,
      "step": 430
    },
    {
      "epoch": 2.72,
      "learning_rate": 4.2997597635544563e-07,
      "loss": 0.2599,
      "step": 431
    },
    {
      "epoch": 2.72,
      "learning_rate": 4.103459726475889e-07,
      "loss": 0.2443,
      "step": 432
    },
    {
      "epoch": 2.73,
      "learning_rate": 3.9116520604279285e-07,
      "loss": 0.2481,
      "step": 433
    },
    {
      "epoch": 2.74,
      "learning_rate": 3.7243457508358784e-07,
      "loss": 0.261,
      "step": 434
    },
    {
      "epoch": 2.74,
      "learning_rate": 3.541549572254488e-07,
      "loss": 0.2734,
      "step": 435
    },
    {
      "epoch": 2.75,
      "learning_rate": 3.3632720879567594e-07,
      "loss": 0.2611,
      "step": 436
    },
    {
      "epoch": 2.75,
      "learning_rate": 3.1895216495329116e-07,
      "loss": 0.2814,
      "step": 437
    },
    {
      "epoch": 2.76,
      "learning_rate": 3.020306396499062e-07,
      "loss": 0.2447,
      "step": 438
    },
    {
      "epoch": 2.77,
      "learning_rate": 2.8556342559159513e-07,
      "loss": 0.2755,
      "step": 439
    },
    {
      "epoch": 2.77,
      "learning_rate": 2.6955129420176193e-07,
      "loss": 0.2661,
      "step": 440
    },
    {
      "epoch": 2.78,
      "learning_rate": 2.539949955849985e-07,
      "loss": 0.2628,
      "step": 441
    },
    {
      "epoch": 2.79,
      "learning_rate": 2.3889525849194573e-07,
      "loss": 0.2727,
      "step": 442
    },
    {
      "epoch": 2.79,
      "learning_rate": 2.2425279028515658e-07,
      "loss": 0.2566,
      "step": 443
    },
    {
      "epoch": 2.8,
      "learning_rate": 2.1006827690595478e-07,
      "loss": 0.2594,
      "step": 444
    },
    {
      "epoch": 2.8,
      "learning_rate": 1.9634238284230945e-07,
      "loss": 0.2515,
      "step": 445
    },
    {
      "epoch": 2.81,
      "learning_rate": 1.830757510976966e-07,
      "loss": 0.2505,
      "step": 446
    },
    {
      "epoch": 2.82,
      "learning_rate": 1.7026900316098217e-07,
      "loss": 0.2476,
      "step": 447
    },
    {
      "epoch": 2.82,
      "learning_rate": 1.5792273897730858e-07,
      "loss": 0.2485,
      "step": 448
    },
    {
      "epoch": 2.83,
      "learning_rate": 1.4603753691998735e-07,
      "loss": 0.2524,
      "step": 449
    },
    {
      "epoch": 2.84,
      "learning_rate": 1.3461395376340502e-07,
      "loss": 0.266,
      "step": 450
    },
    {
      "epoch": 2.84,
      "learning_rate": 1.2365252465694088e-07,
      "loss": 0.2368,
      "step": 451
    },
    {
      "epoch": 2.85,
      "learning_rate": 1.13153763099898e-07,
      "loss": 0.2551,
      "step": 452
    },
    {
      "epoch": 2.86,
      "learning_rate": 1.0311816091744698e-07,
      "loss": 0.2674,
      "step": 453
    },
    {
      "epoch": 2.86,
      "learning_rate": 9.354618823758654e-08,
      "loss": 0.265,
      "step": 454
    },
    {
      "epoch": 2.87,
      "learning_rate": 8.443829346911792e-08,
      "loss": 0.2524,
      "step": 455
    },
    {
      "epoch": 2.87,
      "learning_rate": 7.579490328064265e-08,
      "loss": 0.259,
      "step": 456
    },
    {
      "epoch": 2.88,
      "learning_rate": 6.761642258056977e-08,
      "loss": 0.2615,
      "step": 457
    },
    {
      "epoch": 2.89,
      "learning_rate": 5.990323449815316e-08,
      "loss": 0.2704,
      "step": 458
    },
    {
      "epoch": 2.89,
      "learning_rate": 5.265570036553813e-08,
      "loss": 0.2507,
      "step": 459
    },
    {
      "epoch": 2.9,
      "learning_rate": 4.58741597008372e-08,
      "loss": 0.2576,
      "step": 460
    },
    {
      "epoch": 2.91,
      "learning_rate": 3.955893019222501e-08,
      "loss": 0.2497,
      "step": 461
    },
    {
      "epoch": 2.91,
      "learning_rate": 3.371030768305583e-08,
      "loss": 0.2613,
      "step": 462
    },
    {
      "epoch": 2.92,
      "learning_rate": 2.8328566158002392e-08,
      "loss": 0.2572,
      "step": 463
    },
    {
      "epoch": 2.92,
      "learning_rate": 2.3413957730226144e-08,
      "loss": 0.2496,
      "step": 464
    },
    {
      "epoch": 2.93,
      "learning_rate": 1.896671262955896e-08,
      "loss": 0.2504,
      "step": 465
    },
    {
      "epoch": 2.94,
      "learning_rate": 1.498703919172506e-08,
      "loss": 0.2648,
      "step": 466
    },
    {
      "epoch": 2.94,
      "learning_rate": 1.147512384857663e-08,
      "loss": 0.2602,
      "step": 467
    },
    {
      "epoch": 2.95,
      "learning_rate": 8.431131119361891e-09,
      "loss": 0.2601,
      "step": 468
    },
    {
      "epoch": 2.96,
      "learning_rate": 5.855203603017945e-09,
      "loss": 0.258,
      "step": 469
    },
    {
      "epoch": 2.96,
      "learning_rate": 3.747461971492783e-09,
      "loss": 0.2455,
      "step": 470
    },
    {
      "epoch": 2.97,
      "learning_rate": 2.108004964086474e-09,
      "loss": 0.2588,
      "step": 471
    },
    {
      "epoch": 2.97,
      "learning_rate": 9.369093828326403e-10,
      "loss": 0.2835,
      "step": 472
    },
    {
      "epoch": 2.98,
      "learning_rate": 2.3423008889467134e-10,
      "loss": 0.2562,
      "step": 473
    },
    {
      "epoch": 2.99,
      "learning_rate": 0.0,
      "loss": 0.2487,
      "step": 474
    },
    {
      "epoch": 2.99,
      "step": 474,
      "total_flos": 3.292528880249733e+17,
      "train_loss": 0.4611950102046069,
      "train_runtime": 31248.3915,
      "train_samples_per_second": 1.95,
      "train_steps_per_second": 0.015
    }
  ],
  "max_steps": 474,
  "num_train_epochs": 3,
  "total_flos": 3.292528880249733e+17,
  "trial_name": null,
  "trial_params": null
}