{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "global_step": 750,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "learning_rate": 0,
      "loss": 1.5272,
      "step": 1
    },
    {
      "epoch": 0.01,
      "learning_rate": 0,
      "loss": 1.4685,
      "step": 2
    },
    {
      "epoch": 0.01,
      "learning_rate": 0,
      "loss": 1.5206,
      "step": 3
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0,
      "loss": 1.4631,
      "step": 4
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.421294589150075e-06,
      "loss": 0.9541,
      "step": 5
    },
    {
      "epoch": 0.02,
      "learning_rate": 7.007586128444221e-06,
      "loss": 0.974,
      "step": 6
    },
    {
      "epoch": 0.03,
      "learning_rate": 8.84258917830015e-06,
      "loss": 0.7967,
      "step": 7
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.0265928122321039e-05,
      "loss": 0.769,
      "step": 8
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.1428880717594295e-05,
      "loss": 0.8381,
      "step": 9
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.2412143126717115e-05,
      "loss": 0.8255,
      "step": 10
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.3263883767450225e-05,
      "loss": 0.79,
      "step": 11
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.4015172256888441e-05,
      "loss": 0.8393,
      "step": 12
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.4687222711471113e-05,
      "loss": 0.6941,
      "step": 13
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.529516629701577e-05,
      "loss": 0.6893,
      "step": 14
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.585017530674437e-05,
      "loss": 0.7529,
      "step": 15
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.6360734103293242e-05,
      "loss": 0.7648,
      "step": 16
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.683343771586719e-05,
      "loss": 0.7482,
      "step": 17
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.7273514250765258e-05,
      "loss": 0.75,
      "step": 18
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.76851783566003e-05,
      "loss": 0.7613,
      "step": 19
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.807187734337212e-05,
      "loss": 0.7992,
      "step": 20
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.8436466846038513e-05,
      "loss": 0.6974,
      "step": 21
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.8781338930289855e-05,
      "loss": 0.7481,
      "step": 22
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.9108517300621187e-05,
      "loss": 0.7528,
      "step": 23
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.9419729255161336e-05,
      "loss": 0.6867,
      "step": 24
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.9716460886165846e-05,
      "loss": 0.7166,
      "step": 25
    },
    {
      "epoch": 0.1,
      "learning_rate": 2e-05,
      "loss": 0.7035,
      "step": 26
    },
    {
      "epoch": 0.11,
      "learning_rate": 2e-05,
      "loss": 0.8703,
      "step": 27
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.9972489683631363e-05,
      "loss": 0.7793,
      "step": 28
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.9944979367262724e-05,
      "loss": 0.7478,
      "step": 29
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.991746905089409e-05,
      "loss": 0.7143,
      "step": 30
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.988995873452545e-05,
      "loss": 0.7491,
      "step": 31
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.986244841815681e-05,
      "loss": 0.7789,
      "step": 32
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.9834938101788172e-05,
      "loss": 0.7516,
      "step": 33
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.9807427785419533e-05,
      "loss": 0.796,
      "step": 34
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.9779917469050895e-05,
      "loss": 0.7726,
      "step": 35
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.975240715268226e-05,
      "loss": 0.6836,
      "step": 36
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.9724896836313617e-05,
      "loss": 0.7613,
      "step": 37
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.9697386519944982e-05,
      "loss": 0.6996,
      "step": 38
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.9669876203576343e-05,
      "loss": 0.6864,
      "step": 39
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.9642365887207704e-05,
      "loss": 0.7336,
      "step": 40
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.9614855570839065e-05,
      "loss": 0.6163,
      "step": 41
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.958734525447043e-05,
      "loss": 0.7216,
      "step": 42
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.9559834938101788e-05,
      "loss": 0.6686,
      "step": 43
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.9532324621733152e-05,
      "loss": 0.7211,
      "step": 44
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.9504814305364514e-05,
      "loss": 0.8343,
      "step": 45
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.9477303988995875e-05,
      "loss": 0.8646,
      "step": 46
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.9449793672627236e-05,
      "loss": 0.8046,
      "step": 47
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.9422283356258597e-05,
      "loss": 0.7184,
      "step": 48
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.9394773039889962e-05,
      "loss": 0.7182,
      "step": 49
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.9367262723521323e-05,
      "loss": 0.8039,
      "step": 50
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.9339752407152684e-05,
      "loss": 0.6945,
      "step": 51
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.9312242090784045e-05,
      "loss": 0.746,
      "step": 52
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.928473177441541e-05,
      "loss": 0.7949,
      "step": 53
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.9257221458046768e-05,
      "loss": 0.7124,
      "step": 54
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.9229711141678132e-05,
      "loss": 0.7368,
      "step": 55
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.920220082530949e-05,
      "loss": 0.7541,
      "step": 56
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.9174690508940855e-05,
      "loss": 0.8161,
      "step": 57
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.9147180192572216e-05,
      "loss": 0.7764,
      "step": 58
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.9119669876203577e-05,
      "loss": 0.7333,
      "step": 59
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.909215955983494e-05,
      "loss": 0.7668,
      "step": 60
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.9064649243466303e-05,
      "loss": 0.6952,
      "step": 61
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.903713892709766e-05,
      "loss": 0.7621,
      "step": 62
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.9009628610729025e-05,
      "loss": 0.7054,
      "step": 63
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.8982118294360387e-05,
      "loss": 0.7501,
      "step": 64
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.8954607977991748e-05,
      "loss": 0.7345,
      "step": 65
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.892709766162311e-05,
      "loss": 0.7633,
      "step": 66
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.889958734525447e-05,
      "loss": 0.7899,
      "step": 67
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.8872077028885835e-05,
      "loss": 0.6917,
      "step": 68
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.8844566712517196e-05,
      "loss": 0.8104,
      "step": 69
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.8817056396148557e-05,
      "loss": 0.7783,
      "step": 70
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.878954607977992e-05,
      "loss": 0.7156,
      "step": 71
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.8762035763411283e-05,
      "loss": 0.6756,
      "step": 72
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.873452544704264e-05,
      "loss": 0.7048,
      "step": 73
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.8707015130674006e-05,
      "loss": 0.8007,
      "step": 74
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.8679504814305367e-05,
      "loss": 0.8125,
      "step": 75
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.8651994497936728e-05,
      "loss": 0.7305,
      "step": 76
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.862448418156809e-05,
      "loss": 0.742,
      "step": 77
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.859697386519945e-05,
      "loss": 0.7163,
      "step": 78
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.856946354883081e-05,
      "loss": 0.7119,
      "step": 79
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.8541953232462176e-05,
      "loss": 0.6405,
      "step": 80
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.8514442916093537e-05,
      "loss": 0.8169,
      "step": 81
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.84869325997249e-05,
      "loss": 0.7386,
      "step": 82
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.845942228335626e-05,
      "loss": 0.8507,
      "step": 83
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.843191196698762e-05,
      "loss": 0.6441,
      "step": 84
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.8404401650618982e-05,
      "loss": 0.7248,
      "step": 85
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.8376891334250347e-05,
      "loss": 0.8051,
      "step": 86
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.8349381017881708e-05,
      "loss": 0.7272,
      "step": 87
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.832187070151307e-05,
      "loss": 0.6879,
      "step": 88
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.829436038514443e-05,
      "loss": 0.7123,
      "step": 89
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.826685006877579e-05,
      "loss": 0.6777,
      "step": 90
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.8239339752407156e-05,
      "loss": 0.6839,
      "step": 91
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.8211829436038514e-05,
      "loss": 0.7528,
      "step": 92
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.818431911966988e-05,
      "loss": 0.734,
      "step": 93
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.815680880330124e-05,
      "loss": 0.6664,
      "step": 94
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.81292984869326e-05,
      "loss": 0.6881,
      "step": 95
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.8101788170563962e-05,
      "loss": 0.7597,
      "step": 96
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.8074277854195327e-05,
      "loss": 0.7656,
      "step": 97
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.8046767537826685e-05,
      "loss": 0.7762,
      "step": 98
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.801925722145805e-05,
      "loss": 0.7182,
      "step": 99
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.799174690508941e-05,
      "loss": 0.7527,
      "step": 100
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.7964236588720772e-05,
      "loss": 0.7343,
      "step": 101
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.7936726272352133e-05,
      "loss": 0.8163,
      "step": 102
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.7909215955983494e-05,
      "loss": 0.721,
      "step": 103
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.7881705639614855e-05,
      "loss": 0.7631,
      "step": 104
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.785419532324622e-05,
      "loss": 0.7249,
      "step": 105
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.782668500687758e-05,
      "loss": 0.6999,
      "step": 106
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.7799174690508942e-05,
      "loss": 0.7409,
      "step": 107
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.7771664374140304e-05,
      "loss": 0.7594,
      "step": 108
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.7744154057771665e-05,
      "loss": 0.7373,
      "step": 109
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.771664374140303e-05,
      "loss": 0.7606,
      "step": 110
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.7689133425034387e-05,
      "loss": 0.7624,
      "step": 111
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.7661623108665752e-05,
      "loss": 0.7627,
      "step": 112
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.7634112792297113e-05,
      "loss": 0.717,
      "step": 113
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.7606602475928474e-05,
      "loss": 0.7427,
      "step": 114
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.7579092159559835e-05,
      "loss": 0.6982,
      "step": 115
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.75515818431912e-05,
      "loss": 0.7009,
      "step": 116
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.7524071526822558e-05,
      "loss": 0.6482,
      "step": 117
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.7496561210453922e-05,
      "loss": 0.6886,
      "step": 118
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.7469050894085284e-05,
      "loss": 0.7181,
      "step": 119
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.7441540577716645e-05,
      "loss": 0.7489,
      "step": 120
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.7414030261348006e-05,
      "loss": 0.7776,
      "step": 121
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.738651994497937e-05,
      "loss": 0.7156,
      "step": 122
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.735900962861073e-05,
      "loss": 0.7206,
      "step": 123
    },
    {
      "epoch": 0.5,
      "learning_rate": 1.7331499312242093e-05,
      "loss": 0.6703,
      "step": 124
    },
    {
      "epoch": 0.5,
      "learning_rate": 1.7303988995873454e-05,
      "loss": 0.7133,
      "step": 125
    },
    {
      "epoch": 0.5,
      "learning_rate": 1.7276478679504815e-05,
      "loss": 0.7194,
      "step": 126
    },
    {
      "epoch": 0.51,
      "learning_rate": 1.7248968363136177e-05,
      "loss": 0.737,
      "step": 127
    },
    {
      "epoch": 0.51,
      "learning_rate": 1.7221458046767538e-05,
      "loss": 0.6835,
      "step": 128
    },
    {
      "epoch": 0.52,
      "learning_rate": 1.7193947730398902e-05,
      "loss": 0.7644,
      "step": 129
    },
    {
      "epoch": 0.52,
      "learning_rate": 1.7166437414030264e-05,
      "loss": 0.7093,
      "step": 130
    },
    {
      "epoch": 0.52,
      "learning_rate": 1.7138927097661625e-05,
      "loss": 0.6793,
      "step": 131
    },
    {
      "epoch": 0.53,
      "learning_rate": 1.7111416781292986e-05,
      "loss": 0.712,
      "step": 132
    },
    {
      "epoch": 0.53,
      "learning_rate": 1.708390646492435e-05,
      "loss": 0.6982,
      "step": 133
    },
    {
      "epoch": 0.54,
      "learning_rate": 1.705639614855571e-05,
      "loss": 0.7428,
      "step": 134
    },
    {
      "epoch": 0.54,
      "learning_rate": 1.7028885832187073e-05,
      "loss": 0.739,
      "step": 135
    },
    {
      "epoch": 0.54,
      "learning_rate": 1.700137551581843e-05,
      "loss": 0.65,
      "step": 136
    },
    {
      "epoch": 0.55,
      "learning_rate": 1.6973865199449796e-05,
      "loss": 0.6436,
      "step": 137
    },
    {
      "epoch": 0.55,
      "learning_rate": 1.6946354883081157e-05,
      "loss": 0.7373,
      "step": 138
    },
    {
      "epoch": 0.56,
      "learning_rate": 1.6918844566712518e-05,
      "loss": 0.6389,
      "step": 139
    },
    {
      "epoch": 0.56,
      "learning_rate": 1.689133425034388e-05,
      "loss": 0.7067,
      "step": 140
    },
    {
      "epoch": 0.56,
      "learning_rate": 1.6863823933975244e-05,
      "loss": 0.7783,
      "step": 141
    },
    {
      "epoch": 0.57,
      "learning_rate": 1.6836313617606605e-05,
      "loss": 0.8668,
      "step": 142
    },
    {
      "epoch": 0.57,
      "learning_rate": 1.6808803301237966e-05,
      "loss": 0.7307,
      "step": 143
    },
    {
      "epoch": 0.58,
      "learning_rate": 1.6781292984869327e-05,
      "loss": 0.719,
      "step": 144
    },
    {
      "epoch": 0.58,
      "learning_rate": 1.675378266850069e-05,
      "loss": 0.8011,
      "step": 145
    },
    {
      "epoch": 0.58,
      "learning_rate": 1.672627235213205e-05,
      "loss": 0.7748,
      "step": 146
    },
    {
      "epoch": 0.59,
      "learning_rate": 1.669876203576341e-05,
      "loss": 0.702,
      "step": 147
    },
    {
      "epoch": 0.59,
      "learning_rate": 1.6671251719394776e-05,
      "loss": 0.638,
      "step": 148
    },
    {
      "epoch": 0.6,
      "learning_rate": 1.6643741403026137e-05,
      "loss": 0.7378,
      "step": 149
    },
    {
      "epoch": 0.6,
      "learning_rate": 1.6616231086657498e-05,
      "loss": 0.6801,
      "step": 150
    },
    {
      "epoch": 0.6,
      "learning_rate": 1.658872077028886e-05,
      "loss": 0.796,
      "step": 151
    },
    {
      "epoch": 0.61,
      "learning_rate": 1.6561210453920224e-05,
      "loss": 0.7391,
      "step": 152
    },
    {
      "epoch": 0.61,
      "learning_rate": 1.653370013755158e-05,
      "loss": 0.7216,
      "step": 153
    },
    {
      "epoch": 0.62,
      "learning_rate": 1.6506189821182946e-05,
      "loss": 0.7182,
      "step": 154
    },
    {
      "epoch": 0.62,
      "learning_rate": 1.6478679504814307e-05,
      "loss": 0.6991,
      "step": 155
    },
    {
      "epoch": 0.62,
      "learning_rate": 1.645116918844567e-05,
      "loss": 0.6555,
      "step": 156
    },
    {
      "epoch": 0.63,
      "learning_rate": 1.642365887207703e-05,
      "loss": 0.6516,
      "step": 157
    },
    {
      "epoch": 0.63,
      "learning_rate": 1.639614855570839e-05,
      "loss": 0.728,
      "step": 158
    },
    {
      "epoch": 0.64,
      "learning_rate": 1.6368638239339752e-05,
      "loss": 0.7357,
      "step": 159
    },
    {
      "epoch": 0.64,
      "learning_rate": 1.6341127922971117e-05,
      "loss": 0.6942,
      "step": 160
    },
    {
      "epoch": 0.64,
      "learning_rate": 1.6313617606602478e-05,
      "loss": 0.7153,
      "step": 161
    },
    {
      "epoch": 0.65,
      "learning_rate": 1.628610729023384e-05,
      "loss": 0.7583,
      "step": 162
    },
    {
      "epoch": 0.65,
      "learning_rate": 1.62585969738652e-05,
      "loss": 0.6917,
      "step": 163
    },
    {
      "epoch": 0.66,
      "learning_rate": 1.6231086657496562e-05,
      "loss": 0.6655,
      "step": 164
    },
    {
      "epoch": 0.66,
      "learning_rate": 1.6203576341127923e-05,
      "loss": 0.7071,
      "step": 165
    },
    {
      "epoch": 0.66,
      "learning_rate": 1.6176066024759288e-05,
      "loss": 0.6545,
      "step": 166
    },
    {
      "epoch": 0.67,
      "learning_rate": 1.614855570839065e-05,
      "loss": 0.663,
      "step": 167
    },
    {
      "epoch": 0.67,
      "learning_rate": 1.612104539202201e-05,
      "loss": 0.7543,
      "step": 168
    },
    {
      "epoch": 0.68,
      "learning_rate": 1.609353507565337e-05,
      "loss": 0.6838,
      "step": 169
    },
    {
      "epoch": 0.68,
      "learning_rate": 1.6066024759284732e-05,
      "loss": 0.6506,
      "step": 170
    },
    {
      "epoch": 0.68,
      "learning_rate": 1.6038514442916097e-05,
      "loss": 0.6964,
      "step": 171
    },
    {
      "epoch": 0.69,
      "learning_rate": 1.6011004126547455e-05,
      "loss": 0.7175,
      "step": 172
    },
    {
      "epoch": 0.69,
      "learning_rate": 1.598349381017882e-05,
      "loss": 0.68,
      "step": 173
    },
    {
      "epoch": 0.7,
      "learning_rate": 1.595598349381018e-05,
      "loss": 0.6289,
      "step": 174
    },
    {
      "epoch": 0.7,
      "learning_rate": 1.5928473177441542e-05,
      "loss": 0.6606,
      "step": 175
    },
    {
      "epoch": 0.7,
      "learning_rate": 1.5900962861072903e-05,
      "loss": 0.7564,
      "step": 176
    },
    {
      "epoch": 0.71,
      "learning_rate": 1.5873452544704268e-05,
      "loss": 0.6964,
      "step": 177
    },
    {
      "epoch": 0.71,
      "learning_rate": 1.5845942228335625e-05,
      "loss": 0.6616,
      "step": 178
    },
    {
      "epoch": 0.72,
      "learning_rate": 1.581843191196699e-05,
      "loss": 0.7258,
      "step": 179
    },
    {
      "epoch": 0.72,
      "learning_rate": 1.579092159559835e-05,
      "loss": 0.7294,
      "step": 180
    },
    {
      "epoch": 0.72,
      "learning_rate": 1.5763411279229712e-05,
      "loss": 0.7013,
      "step": 181
    },
    {
      "epoch": 0.73,
      "learning_rate": 1.5735900962861074e-05,
      "loss": 0.6723,
      "step": 182
    },
    {
      "epoch": 0.73,
      "learning_rate": 1.5708390646492435e-05,
      "loss": 0.6372,
      "step": 183
    },
    {
      "epoch": 0.74,
      "learning_rate": 1.5680880330123796e-05,
      "loss": 0.7439,
      "step": 184
    },
    {
      "epoch": 0.74,
      "learning_rate": 1.565337001375516e-05,
      "loss": 0.6167,
      "step": 185
    },
    {
      "epoch": 0.74,
      "learning_rate": 1.5625859697386522e-05,
      "loss": 0.6366,
      "step": 186
    },
    {
      "epoch": 0.75,
      "learning_rate": 1.5598349381017883e-05,
      "loss": 0.7574,
      "step": 187
    },
    {
      "epoch": 0.75,
      "learning_rate": 1.5570839064649244e-05,
      "loss": 0.7614,
      "step": 188
    },
    {
      "epoch": 0.76,
      "learning_rate": 1.5543328748280606e-05,
      "loss": 0.7012,
      "step": 189
    },
    {
      "epoch": 0.76,
      "learning_rate": 1.551581843191197e-05,
      "loss": 0.65,
      "step": 190
    },
    {
      "epoch": 0.76,
      "learning_rate": 1.5488308115543328e-05,
      "loss": 0.7321,
      "step": 191
    },
    {
      "epoch": 0.77,
      "learning_rate": 1.5460797799174693e-05,
      "loss": 0.6749,
      "step": 192
    },
    {
      "epoch": 0.77,
      "learning_rate": 1.5433287482806054e-05,
      "loss": 0.7115,
      "step": 193
    },
    {
      "epoch": 0.78,
      "learning_rate": 1.5405777166437415e-05,
      "loss": 0.7541,
      "step": 194
    },
    {
      "epoch": 0.78,
      "learning_rate": 1.5378266850068776e-05,
      "loss": 0.7499,
      "step": 195
    },
    {
      "epoch": 0.78,
      "learning_rate": 1.535075653370014e-05,
      "loss": 0.7071,
      "step": 196
    },
    {
      "epoch": 0.79,
      "learning_rate": 1.53232462173315e-05,
      "loss": 0.7087,
      "step": 197
    },
    {
      "epoch": 0.79,
      "learning_rate": 1.5295735900962863e-05,
      "loss": 0.7065,
      "step": 198
    },
    {
      "epoch": 0.8,
      "learning_rate": 1.5268225584594224e-05,
      "loss": 0.7079,
      "step": 199
    },
    {
      "epoch": 0.8,
      "learning_rate": 1.5240715268225586e-05,
      "loss": 0.7535,
      "step": 200
    },
    {
      "epoch": 0.8,
      "learning_rate": 1.5213204951856948e-05,
      "loss": 0.7611,
      "step": 201
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.5185694635488308e-05,
      "loss": 0.7247,
      "step": 202
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.5158184319119671e-05,
      "loss": 0.7209,
      "step": 203
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.5130674002751034e-05,
      "loss": 0.7039,
      "step": 204
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.5103163686382393e-05,
      "loss": 0.7522,
      "step": 205
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.5075653370013756e-05,
      "loss": 0.6941,
      "step": 206
    },
    {
      "epoch": 0.83,
      "learning_rate": 1.5048143053645119e-05,
      "loss": 0.6611,
      "step": 207
    },
    {
      "epoch": 0.83,
      "learning_rate": 1.5020632737276479e-05,
      "loss": 0.7778,
      "step": 208
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.4993122420907842e-05,
      "loss": 0.6705,
      "step": 209
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.4965612104539204e-05,
      "loss": 0.6179,
      "step": 210
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.4938101788170564e-05,
      "loss": 0.675,
      "step": 211
    },
    {
      "epoch": 0.85,
      "learning_rate": 1.4910591471801927e-05,
      "loss": 0.6672,
      "step": 212
    },
    {
      "epoch": 0.85,
      "learning_rate": 1.4883081155433288e-05,
      "loss": 0.7322,
      "step": 213
    },
    {
      "epoch": 0.86,
      "learning_rate": 1.4855570839064651e-05,
      "loss": 0.7011,
      "step": 214
    },
    {
      "epoch": 0.86,
      "learning_rate": 1.4828060522696012e-05,
      "loss": 0.7106,
      "step": 215
    },
    {
      "epoch": 0.86,
      "learning_rate": 1.4800550206327373e-05,
      "loss": 0.7264,
      "step": 216
    },
    {
      "epoch": 0.87,
      "learning_rate": 1.4773039889958736e-05,
      "loss": 0.7744,
      "step": 217
    },
    {
      "epoch": 0.87,
      "learning_rate": 1.4745529573590097e-05,
      "loss": 0.684,
      "step": 218
    },
    {
      "epoch": 0.88,
      "learning_rate": 1.4718019257221459e-05,
      "loss": 0.6544,
      "step": 219
    },
    {
      "epoch": 0.88,
      "learning_rate": 1.4690508940852822e-05,
      "loss": 0.7054,
      "step": 220
    },
    {
      "epoch": 0.88,
      "learning_rate": 1.4662998624484185e-05,
      "loss": 0.6471,
      "step": 221
    },
    {
      "epoch": 0.89,
      "learning_rate": 1.4635488308115544e-05,
      "loss": 0.6562,
      "step": 222
    },
    {
      "epoch": 0.89,
      "learning_rate": 1.4607977991746907e-05,
      "loss": 0.7364,
      "step": 223
    },
    {
      "epoch": 0.9,
      "learning_rate": 1.4580467675378266e-05,
      "loss": 0.752,
      "step": 224
    },
    {
      "epoch": 0.9,
      "learning_rate": 1.455295735900963e-05,
      "loss": 0.6894,
      "step": 225
    },
    {
      "epoch": 0.9,
      "learning_rate": 1.4525447042640992e-05,
      "loss": 0.7072,
      "step": 226
    },
    {
      "epoch": 0.91,
      "learning_rate": 1.4497936726272352e-05,
      "loss": 0.6649,
      "step": 227
    },
    {
      "epoch": 0.91,
      "learning_rate": 1.4470426409903715e-05,
      "loss": 0.7411,
      "step": 228
    },
    {
      "epoch": 0.92,
      "learning_rate": 1.4442916093535078e-05,
      "loss": 0.7079,
      "step": 229
    },
    {
      "epoch": 0.92,
      "learning_rate": 1.4415405777166439e-05,
      "loss": 0.6625,
      "step": 230
    },
    {
      "epoch": 0.92,
      "learning_rate": 1.43878954607978e-05,
      "loss": 0.7616,
      "step": 231
    },
    {
      "epoch": 0.93,
      "learning_rate": 1.4360385144429163e-05,
      "loss": 0.7437,
      "step": 232
    },
    {
      "epoch": 0.93,
      "learning_rate": 1.4332874828060524e-05,
      "loss": 0.7377,
      "step": 233
    },
    {
      "epoch": 0.94,
      "learning_rate": 1.4305364511691885e-05,
      "loss": 0.6102,
      "step": 234
    },
    {
      "epoch": 0.94,
      "learning_rate": 1.4277854195323247e-05,
      "loss": 0.7094,
      "step": 235
    },
    {
      "epoch": 0.94,
      "learning_rate": 1.425034387895461e-05,
      "loss": 0.694,
      "step": 236
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.4222833562585972e-05,
      "loss": 0.6074,
      "step": 237
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.4195323246217332e-05,
      "loss": 0.7402,
      "step": 238
    },
    {
      "epoch": 0.96,
      "learning_rate": 1.4167812929848695e-05,
      "loss": 0.6544,
      "step": 239
    },
    {
      "epoch": 0.96,
      "learning_rate": 1.4140302613480058e-05,
      "loss": 0.736,
      "step": 240
    },
    {
      "epoch": 0.96,
      "learning_rate": 1.4112792297111417e-05,
      "loss": 0.8055,
      "step": 241
    },
    {
      "epoch": 0.97,
      "learning_rate": 1.408528198074278e-05,
      "loss": 0.7056,
      "step": 242
    },
    {
      "epoch": 0.97,
      "learning_rate": 1.4057771664374143e-05,
      "loss": 0.7534,
      "step": 243
    },
    {
      "epoch": 0.98,
      "learning_rate": 1.4030261348005502e-05,
      "loss": 0.6616,
      "step": 244
    },
    {
      "epoch": 0.98,
      "learning_rate": 1.4002751031636865e-05,
      "loss": 0.6677,
      "step": 245
    },
    {
      "epoch": 0.98,
      "learning_rate": 1.3975240715268225e-05,
      "loss": 0.6832,
      "step": 246
    },
    {
      "epoch": 0.99,
      "learning_rate": 1.3947730398899588e-05,
      "loss": 0.68,
      "step": 247
    },
    {
      "epoch": 0.99,
      "learning_rate": 1.392022008253095e-05,
      "loss": 0.6881,
      "step": 248
    },
    {
      "epoch": 1.0,
      "learning_rate": 1.3892709766162312e-05,
      "loss": 0.6452,
      "step": 249
    },
    {
      "epoch": 1.0,
      "learning_rate": 1.3865199449793673e-05,
      "loss": 0.6573,
      "step": 250
    },
    {
      "epoch": 1.0,
      "learning_rate": 1.3837689133425036e-05,
      "loss": 0.3832,
      "step": 251
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.3810178817056397e-05,
      "loss": 0.4213,
      "step": 252
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.3782668500687758e-05,
      "loss": 0.3994,
      "step": 253
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.3755158184319121e-05,
      "loss": 0.3177,
      "step": 254
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.3727647867950483e-05,
      "loss": 0.3202,
      "step": 255
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.3700137551581845e-05,
      "loss": 0.3,
      "step": 256
    },
    {
      "epoch": 1.03,
      "learning_rate": 1.3672627235213205e-05,
      "loss": 0.3263,
      "step": 257
    },
    {
      "epoch": 1.03,
      "learning_rate": 1.3645116918844568e-05,
      "loss": 0.3288,
      "step": 258
    },
    {
      "epoch": 1.04,
      "learning_rate": 1.361760660247593e-05,
      "loss": 0.3797,
      "step": 259
    },
    {
      "epoch": 1.04,
      "learning_rate": 1.359009628610729e-05,
      "loss": 0.3363,
      "step": 260
    },
    {
      "epoch": 1.04,
      "learning_rate": 1.3562585969738653e-05,
      "loss": 0.3725,
      "step": 261
    },
    {
      "epoch": 1.05,
      "learning_rate": 1.3535075653370016e-05,
      "loss": 0.3074,
      "step": 262
    },
    {
      "epoch": 1.05,
      "learning_rate": 1.3507565337001376e-05,
      "loss": 0.3079,
      "step": 263
    },
    {
      "epoch": 1.06,
      "learning_rate": 1.3480055020632738e-05,
      "loss": 0.2891,
      "step": 264
    },
    {
      "epoch": 1.06,
      "learning_rate": 1.3452544704264101e-05,
      "loss": 0.3187,
      "step": 265
    },
    {
      "epoch": 1.06,
      "learning_rate": 1.3425034387895461e-05,
      "loss": 0.2769,
      "step": 266
    },
    {
      "epoch": 1.07,
      "learning_rate": 1.3397524071526824e-05,
      "loss": 0.2724,
      "step": 267
    },
    {
      "epoch": 1.07,
      "learning_rate": 1.3370013755158185e-05,
      "loss": 0.3177,
      "step": 268
    },
    {
      "epoch": 1.08,
      "learning_rate": 1.3342503438789546e-05,
      "loss": 0.2724,
      "step": 269
    },
    {
      "epoch": 1.08,
      "learning_rate": 1.3314993122420909e-05,
      "loss": 0.2723,
      "step": 270
    },
    {
      "epoch": 1.08,
      "learning_rate": 1.328748280605227e-05,
      "loss": 0.3022,
      "step": 271
    },
    {
      "epoch": 1.09,
      "learning_rate": 1.3259972489683632e-05,
      "loss": 0.284,
      "step": 272
    },
    {
      "epoch": 1.09,
      "learning_rate": 1.3232462173314994e-05,
      "loss": 0.304,
      "step": 273
    },
    {
      "epoch": 1.1,
      "learning_rate": 1.3204951856946356e-05,
      "loss": 0.308,
      "step": 274
    },
    {
      "epoch": 1.1,
      "learning_rate": 1.3177441540577719e-05,
      "loss": 0.3128,
      "step": 275
    },
    {
      "epoch": 1.1,
      "learning_rate": 1.314993122420908e-05,
      "loss": 0.3224,
      "step": 276
    },
    {
      "epoch": 1.11,
      "learning_rate": 1.3122420907840441e-05,
      "loss": 0.3139,
      "step": 277
    },
    {
      "epoch": 1.11,
      "learning_rate": 1.3094910591471804e-05,
      "loss": 0.3198,
      "step": 278
    },
    {
      "epoch": 1.12,
      "learning_rate": 1.3067400275103163e-05,
      "loss": 0.2906,
      "step": 279
    },
    {
      "epoch": 1.12,
      "learning_rate": 1.3039889958734526e-05,
      "loss": 0.3141,
      "step": 280
    },
    {
      "epoch": 1.12,
      "learning_rate": 1.301237964236589e-05,
      "loss": 0.3022,
      "step": 281
    },
    {
      "epoch": 1.13,
      "learning_rate": 1.2984869325997249e-05,
      "loss": 0.3306,
      "step": 282
    },
    {
      "epoch": 1.13,
      "learning_rate": 1.2957359009628612e-05,
      "loss": 0.2575,
      "step": 283
    },
    {
      "epoch": 1.14,
      "learning_rate": 1.2929848693259975e-05,
      "loss": 0.3409,
      "step": 284
    },
    {
      "epoch": 1.14,
      "learning_rate": 1.2902338376891334e-05,
      "loss": 0.2825,
      "step": 285
    },
    {
      "epoch": 1.14,
      "learning_rate": 1.2874828060522697e-05,
      "loss": 0.3299,
      "step": 286
    },
    {
      "epoch": 1.15,
      "learning_rate": 1.284731774415406e-05,
      "loss": 0.2843,
      "step": 287
    },
    {
      "epoch": 1.15,
      "learning_rate": 1.281980742778542e-05,
      "loss": 0.3134,
      "step": 288
    },
    {
      "epoch": 1.16,
      "learning_rate": 1.2792297111416782e-05,
      "loss": 0.276,
      "step": 289
    },
    {
      "epoch": 1.16,
      "learning_rate": 1.2764786795048143e-05,
      "loss": 0.3417,
      "step": 290
    },
    {
      "epoch": 1.16,
      "learning_rate": 1.2737276478679506e-05,
      "loss": 0.2623,
      "step": 291
    },
    {
      "epoch": 1.17,
      "learning_rate": 1.2709766162310868e-05,
      "loss": 0.2778,
      "step": 292
    },
    {
      "epoch": 1.17,
      "learning_rate": 1.2682255845942229e-05,
      "loss": 0.3325,
      "step": 293
    },
    {
      "epoch": 1.18,
      "learning_rate": 1.2654745529573592e-05,
      "loss": 0.3026,
      "step": 294
    },
    {
      "epoch": 1.18,
      "learning_rate": 1.2627235213204953e-05,
      "loss": 0.3896,
      "step": 295
    },
    {
      "epoch": 1.18,
      "learning_rate": 1.2599724896836314e-05,
      "loss": 0.3385,
      "step": 296
    },
    {
      "epoch": 1.19,
      "learning_rate": 1.2572214580467677e-05,
      "loss": 0.3296,
      "step": 297
    },
    {
      "epoch": 1.19,
      "learning_rate": 1.254470426409904e-05,
      "loss": 0.2828,
      "step": 298
    },
    {
      "epoch": 1.2,
      "learning_rate": 1.25171939477304e-05,
      "loss": 0.3029,
      "step": 299
    },
    {
      "epoch": 1.2,
      "learning_rate": 1.2489683631361762e-05,
      "loss": 0.3288,
      "step": 300
    },
    {
      "epoch": 1.2,
      "learning_rate": 1.2462173314993125e-05,
      "loss": 0.2971,
      "step": 301
    },
    {
      "epoch": 1.21,
      "learning_rate": 1.2434662998624485e-05,
      "loss": 0.2642,
      "step": 302
    },
    {
      "epoch": 1.21,
      "learning_rate": 1.2407152682255848e-05,
      "loss": 0.2711,
      "step": 303
    },
    {
      "epoch": 1.22,
      "learning_rate": 1.2379642365887207e-05,
      "loss": 0.3037,
      "step": 304
    },
    {
      "epoch": 1.22,
      "learning_rate": 1.235213204951857e-05,
      "loss": 0.2917,
      "step": 305
    },
    {
      "epoch": 1.22,
      "learning_rate": 1.2324621733149933e-05,
      "loss": 0.32,
      "step": 306
    },
    {
      "epoch": 1.23,
      "learning_rate": 1.2297111416781292e-05,
      "loss": 0.2759,
      "step": 307
    },
    {
      "epoch": 1.23,
      "learning_rate": 1.2269601100412655e-05,
      "loss": 0.3036,
      "step": 308
    },
    {
      "epoch": 1.24,
      "learning_rate": 1.2242090784044018e-05,
      "loss": 0.3284,
      "step": 309
    },
    {
      "epoch": 1.24,
      "learning_rate": 1.221458046767538e-05,
      "loss": 0.2982,
      "step": 310
    },
    {
      "epoch": 1.24,
      "learning_rate": 1.218707015130674e-05,
      "loss": 0.3157,
      "step": 311
    },
    {
      "epoch": 1.25,
      "learning_rate": 1.2159559834938104e-05,
      "loss": 0.316,
      "step": 312
    },
    {
      "epoch": 1.25,
      "learning_rate": 1.2132049518569465e-05,
      "loss": 0.3031,
      "step": 313
    },
    {
      "epoch": 1.26,
      "learning_rate": 1.2104539202200826e-05,
      "loss": 0.3152,
      "step": 314
    },
    {
      "epoch": 1.26,
      "learning_rate": 1.2077028885832187e-05,
      "loss": 0.2929,
      "step": 315
    },
    {
      "epoch": 1.26,
      "learning_rate": 1.204951856946355e-05,
      "loss": 0.2616,
      "step": 316
    },
    {
      "epoch": 1.27,
      "learning_rate": 1.2022008253094913e-05,
      "loss": 0.3035,
      "step": 317
    },
    {
      "epoch": 1.27,
      "learning_rate": 1.1994497936726273e-05,
      "loss": 0.2955,
      "step": 318
    },
    {
      "epoch": 1.28,
      "learning_rate": 1.1966987620357635e-05,
      "loss": 0.2996,
      "step": 319
    },
    {
      "epoch": 1.28,
      "learning_rate": 1.1939477303988998e-05,
      "loss": 0.3157,
      "step": 320
    },
    {
      "epoch": 1.28,
      "learning_rate": 1.1911966987620358e-05,
      "loss": 0.3141,
      "step": 321
    },
    {
      "epoch": 1.29,
      "learning_rate": 1.188445667125172e-05,
      "loss": 0.3236,
      "step": 322
    },
    {
      "epoch": 1.29,
      "learning_rate": 1.1856946354883084e-05,
      "loss": 0.3105,
      "step": 323
    },
    {
      "epoch": 1.3,
      "learning_rate": 1.1829436038514443e-05,
      "loss": 0.3012,
      "step": 324
    },
    {
      "epoch": 1.3,
      "learning_rate": 1.1801925722145806e-05,
      "loss": 0.2917,
      "step": 325
    },
    {
      "epoch": 1.3,
      "learning_rate": 1.1774415405777167e-05,
      "loss": 0.3089,
      "step": 326
    },
    {
      "epoch": 1.31,
      "learning_rate": 1.1746905089408529e-05,
      "loss": 0.3057,
      "step": 327
    },
    {
      "epoch": 1.31,
      "learning_rate": 1.1719394773039891e-05,
      "loss": 0.314,
      "step": 328
    },
    {
      "epoch": 1.32,
      "learning_rate": 1.1691884456671253e-05,
      "loss": 0.3231,
      "step": 329
    },
    {
      "epoch": 1.32,
      "learning_rate": 1.1664374140302614e-05,
      "loss": 0.3227,
      "step": 330
    },
    {
      "epoch": 1.32,
      "learning_rate": 1.1636863823933977e-05,
      "loss": 0.2869,
      "step": 331
    },
    {
      "epoch": 1.33,
      "learning_rate": 1.1609353507565338e-05,
      "loss": 0.3323,
      "step": 332
    },
    {
      "epoch": 1.33,
      "learning_rate": 1.1581843191196699e-05,
      "loss": 0.264,
      "step": 333
    },
    {
      "epoch": 1.34,
      "learning_rate": 1.1554332874828062e-05,
      "loss": 0.2834,
      "step": 334
    },
    {
      "epoch": 1.34,
      "learning_rate": 1.1526822558459423e-05,
      "loss": 0.3327,
      "step": 335
    },
    {
      "epoch": 1.34,
      "learning_rate": 1.1499312242090786e-05,
      "loss": 0.3061,
      "step": 336
    },
    {
      "epoch": 1.35,
      "learning_rate": 1.1471801925722146e-05,
      "loss": 0.29,
      "step": 337
    },
    {
      "epoch": 1.35,
      "learning_rate": 1.1444291609353509e-05,
      "loss": 0.3232,
      "step": 338
    },
    {
      "epoch": 1.36,
      "learning_rate": 1.1416781292984871e-05,
      "loss": 0.3349,
      "step": 339
    },
    {
      "epoch": 1.36,
      "learning_rate": 1.1389270976616231e-05,
      "loss": 0.2967,
      "step": 340
    },
    {
      "epoch": 1.36,
      "learning_rate": 1.1361760660247594e-05,
      "loss": 0.322,
      "step": 341
    },
    {
      "epoch": 1.37,
      "learning_rate": 1.1334250343878957e-05,
      "loss": 0.3212,
      "step": 342
    },
    {
      "epoch": 1.37,
      "learning_rate": 1.1306740027510316e-05,
      "loss": 0.3236,
      "step": 343
    },
    {
      "epoch": 1.38,
      "learning_rate": 1.127922971114168e-05,
      "loss": 0.3043,
      "step": 344
    },
    {
      "epoch": 1.38,
      "learning_rate": 1.1251719394773042e-05,
      "loss": 0.3347,
      "step": 345
    },
    {
      "epoch": 1.38,
      "learning_rate": 1.1224209078404402e-05,
      "loss": 0.335,
      "step": 346
    },
    {
      "epoch": 1.39,
      "learning_rate": 1.1196698762035765e-05,
      "loss": 0.3274,
      "step": 347
    },
    {
      "epoch": 1.39,
      "learning_rate": 1.1169188445667126e-05,
      "loss": 0.2917,
      "step": 348
    },
    {
      "epoch": 1.4,
      "learning_rate": 1.1141678129298487e-05,
      "loss": 0.3259,
      "step": 349
    },
    {
      "epoch": 1.4,
      "learning_rate": 1.111416781292985e-05,
      "loss": 0.2635,
      "step": 350
    },
    {
      "epoch": 1.4,
      "learning_rate": 1.1086657496561211e-05,
      "loss": 0.2856,
      "step": 351
    },
    {
      "epoch": 1.41,
      "learning_rate": 1.1059147180192574e-05,
      "loss": 0.329,
      "step": 352
    },
    {
      "epoch": 1.41,
      "learning_rate": 1.1031636863823935e-05,
      "loss": 0.31,
      "step": 353
    },
    {
      "epoch": 1.42,
      "learning_rate": 1.1004126547455296e-05,
      "loss": 0.2703,
      "step": 354
    },
    {
      "epoch": 1.42,
      "learning_rate": 1.097661623108666e-05,
      "loss": 0.3278,
      "step": 355
    },
    {
      "epoch": 1.42,
      "learning_rate": 1.094910591471802e-05,
      "loss": 0.3352,
      "step": 356
    },
    {
      "epoch": 1.43,
      "learning_rate": 1.0921595598349382e-05,
      "loss": 0.2798,
      "step": 357
    },
    {
      "epoch": 1.43,
      "learning_rate": 1.0894085281980745e-05,
      "loss": 0.3067,
      "step": 358
    },
    {
      "epoch": 1.44,
      "learning_rate": 1.0866574965612104e-05,
      "loss": 0.2838,
      "step": 359
    },
    {
      "epoch": 1.44,
      "learning_rate": 1.0839064649243467e-05,
      "loss": 0.3022,
      "step": 360
    },
    {
      "epoch": 1.44,
      "learning_rate": 1.081155433287483e-05,
      "loss": 0.3093,
      "step": 361
    },
    {
      "epoch": 1.45,
      "learning_rate": 1.078404401650619e-05,
      "loss": 0.3349,
      "step": 362
    },
    {
      "epoch": 1.45,
      "learning_rate": 1.0756533700137552e-05,
      "loss": 0.307,
      "step": 363
    },
    {
      "epoch": 1.46,
      "learning_rate": 1.0729023383768915e-05,
      "loss": 0.3127,
      "step": 364
    },
    {
      "epoch": 1.46,
      "learning_rate": 1.0701513067400275e-05,
      "loss": 0.2595,
      "step": 365
    },
    {
      "epoch": 1.46,
      "learning_rate": 1.0674002751031638e-05,
      "loss": 0.2923,
      "step": 366
    },
    {
      "epoch": 1.47,
      "learning_rate": 1.0646492434663e-05,
      "loss": 0.2882,
      "step": 367
    },
    {
      "epoch": 1.47,
      "learning_rate": 1.061898211829436e-05,
      "loss": 0.2729,
      "step": 368
    },
    {
      "epoch": 1.48,
      "learning_rate": 1.0591471801925723e-05,
      "loss": 0.3035,
      "step": 369
    },
    {
      "epoch": 1.48,
      "learning_rate": 1.0563961485557084e-05,
      "loss": 0.3469,
      "step": 370
    },
    {
      "epoch": 1.48,
      "learning_rate": 1.0536451169188447e-05,
      "loss": 0.329,
      "step": 371
    },
    {
      "epoch": 1.49,
      "learning_rate": 1.0508940852819808e-05,
      "loss": 0.3302,
      "step": 372
    },
    {
      "epoch": 1.49,
      "learning_rate": 1.048143053645117e-05,
      "loss": 0.3358,
      "step": 373
    },
    {
      "epoch": 1.5,
      "learning_rate": 1.0453920220082532e-05,
      "loss": 0.3315,
      "step": 374
    },
    {
      "epoch": 1.5,
      "learning_rate": 1.0426409903713894e-05,
      "loss": 0.2801,
      "step": 375
    },
    {
      "epoch": 1.5,
      "learning_rate": 1.0398899587345255e-05,
      "loss": 0.2818,
      "step": 376
    },
    {
      "epoch": 1.51,
      "learning_rate": 1.0371389270976618e-05,
      "loss": 0.2859,
      "step": 377
    },
    {
      "epoch": 1.51,
      "learning_rate": 1.034387895460798e-05,
      "loss": 0.355,
      "step": 378
    },
    {
      "epoch": 1.52,
      "learning_rate": 1.031636863823934e-05,
      "loss": 0.2643,
      "step": 379
    },
    {
      "epoch": 1.52,
      "learning_rate": 1.0288858321870703e-05,
      "loss": 0.257,
      "step": 380
    },
    {
      "epoch": 1.52,
      "learning_rate": 1.0261348005502063e-05,
      "loss": 0.354,
      "step": 381
    },
    {
      "epoch": 1.53,
      "learning_rate": 1.0233837689133425e-05,
      "loss": 0.3119,
      "step": 382
    },
    {
      "epoch": 1.53,
      "learning_rate": 1.0206327372764788e-05,
      "loss": 0.3228,
      "step": 383
    },
    {
      "epoch": 1.54,
      "learning_rate": 1.0178817056396148e-05,
      "loss": 0.3063,
      "step": 384
    },
    {
      "epoch": 1.54,
      "learning_rate": 1.015130674002751e-05,
      "loss": 0.3218,
      "step": 385
    },
    {
      "epoch": 1.54,
      "learning_rate": 1.0123796423658874e-05,
      "loss": 0.3326,
      "step": 386
    },
    {
      "epoch": 1.55,
      "learning_rate": 1.0096286107290235e-05,
      "loss": 0.3411,
      "step": 387
    },
    {
      "epoch": 1.55,
      "learning_rate": 1.0068775790921596e-05,
      "loss": 0.3243,
      "step": 388
    },
    {
      "epoch": 1.56,
      "learning_rate": 1.0041265474552959e-05,
      "loss": 0.349,
      "step": 389
    },
    {
      "epoch": 1.56,
      "learning_rate": 1.001375515818432e-05,
      "loss": 0.2975,
      "step": 390
    },
    {
      "epoch": 1.56,
      "learning_rate": 9.986244841815681e-06,
      "loss": 0.3078,
      "step": 391
    },
    {
      "epoch": 1.57,
      "learning_rate": 9.958734525447044e-06,
      "loss": 0.3176,
      "step": 392
    },
    {
      "epoch": 1.57,
      "learning_rate": 9.931224209078406e-06,
      "loss": 0.2235,
      "step": 393
    },
    {
      "epoch": 1.58,
      "learning_rate": 9.903713892709767e-06,
      "loss": 0.3255,
      "step": 394
    },
    {
      "epoch": 1.58,
      "learning_rate": 9.87620357634113e-06,
      "loss": 0.3365,
      "step": 395
    },
    {
      "epoch": 1.58,
      "learning_rate": 9.848693259972491e-06,
      "loss": 0.2892,
      "step": 396
    },
    {
      "epoch": 1.59,
      "learning_rate": 9.821182943603852e-06,
      "loss": 0.3018,
      "step": 397
    },
    {
      "epoch": 1.59,
      "learning_rate": 9.793672627235215e-06,
      "loss": 0.2908,
      "step": 398
    },
    {
      "epoch": 1.6,
      "learning_rate": 9.766162310866576e-06,
      "loss": 0.2969,
      "step": 399
    },
    {
      "epoch": 1.6,
      "learning_rate": 9.738651994497937e-06,
      "loss": 0.24,
      "step": 400
    },
    {
      "epoch": 1.6,
      "learning_rate": 9.711141678129299e-06,
      "loss": 0.2877,
      "step": 401
    },
    {
      "epoch": 1.61,
      "learning_rate": 9.683631361760661e-06,
      "loss": 0.2665,
      "step": 402
    },
    {
      "epoch": 1.61,
      "learning_rate": 9.656121045392023e-06,
      "loss": 0.3016,
      "step": 403
    },
    {
      "epoch": 1.62,
      "learning_rate": 9.628610729023384e-06,
      "loss": 0.3647,
      "step": 404
    },
    {
      "epoch": 1.62,
      "learning_rate": 9.601100412654745e-06,
      "loss": 0.2788,
      "step": 405
    },
    {
      "epoch": 1.62,
      "learning_rate": 9.573590096286108e-06,
      "loss": 0.2821,
      "step": 406
    },
    {
      "epoch": 1.63,
      "learning_rate": 9.54607977991747e-06,
      "loss": 0.2481,
      "step": 407
    },
    {
      "epoch": 1.63,
      "learning_rate": 9.51856946354883e-06,
      "loss": 0.3071,
      "step": 408
    },
    {
      "epoch": 1.64,
      "learning_rate": 9.491059147180193e-06,
      "loss": 0.2887,
      "step": 409
    },
    {
      "epoch": 1.64,
      "learning_rate": 9.463548830811555e-06,
      "loss": 0.3086,
      "step": 410
    },
    {
      "epoch": 1.64,
      "learning_rate": 9.436038514442917e-06,
      "loss": 0.3034,
      "step": 411
    },
    {
      "epoch": 1.65,
      "learning_rate": 9.408528198074279e-06,
      "loss": 0.3142,
      "step": 412
    },
    {
      "epoch": 1.65,
      "learning_rate": 9.381017881705642e-06,
      "loss": 0.2837,
      "step": 413
    },
    {
      "epoch": 1.66,
      "learning_rate": 9.353507565337003e-06,
      "loss": 0.2937,
      "step": 414
    },
    {
      "epoch": 1.66,
      "learning_rate": 9.325997248968364e-06,
      "loss": 0.3049,
      "step": 415
    },
    {
      "epoch": 1.66,
      "learning_rate": 9.298486932599725e-06,
      "loss": 0.2905,
      "step": 416
    },
    {
      "epoch": 1.67,
      "learning_rate": 9.270976616231088e-06,
      "loss": 0.3221,
      "step": 417
    },
    {
      "epoch": 1.67,
      "learning_rate": 9.24346629986245e-06,
      "loss": 0.3128,
      "step": 418
    },
    {
      "epoch": 1.68,
      "learning_rate": 9.21595598349381e-06,
      "loss": 0.318,
      "step": 419
    },
    {
      "epoch": 1.68,
      "learning_rate": 9.188445667125173e-06,
      "loss": 0.2852,
      "step": 420
    },
    {
      "epoch": 1.68,
      "learning_rate": 9.160935350756535e-06,
      "loss": 0.2334,
      "step": 421
    },
    {
      "epoch": 1.69,
      "learning_rate": 9.133425034387896e-06,
      "loss": 0.3457,
      "step": 422
    },
    {
      "epoch": 1.69,
      "learning_rate": 9.105914718019257e-06,
      "loss": 0.2861,
      "step": 423
    },
    {
      "epoch": 1.7,
      "learning_rate": 9.07840440165062e-06,
      "loss": 0.2817,
      "step": 424
    },
    {
      "epoch": 1.7,
      "learning_rate": 9.050894085281981e-06,
      "loss": 0.3062,
      "step": 425
    },
    {
      "epoch": 1.7,
      "learning_rate": 9.023383768913342e-06,
      "loss": 0.3121,
      "step": 426
    },
    {
      "epoch": 1.71,
      "learning_rate": 8.995873452544705e-06,
      "loss": 0.3047,
      "step": 427
    },
    {
      "epoch": 1.71,
      "learning_rate": 8.968363136176066e-06,
      "loss": 0.3511,
      "step": 428
    },
    {
      "epoch": 1.72,
      "learning_rate": 8.940852819807428e-06,
      "loss": 0.284,
      "step": 429
    },
    {
      "epoch": 1.72,
      "learning_rate": 8.91334250343879e-06,
      "loss": 0.2439,
      "step": 430
    },
    {
      "epoch": 1.72,
      "learning_rate": 8.885832187070152e-06,
      "loss": 0.2665,
      "step": 431
    },
    {
      "epoch": 1.73,
      "learning_rate": 8.858321870701515e-06,
      "loss": 0.3014,
      "step": 432
    },
    {
      "epoch": 1.73,
      "learning_rate": 8.830811554332876e-06,
      "loss": 0.2791,
      "step": 433
    },
    {
      "epoch": 1.74,
      "learning_rate": 8.803301237964237e-06,
      "loss": 0.3252,
      "step": 434
    },
    {
      "epoch": 1.74,
      "learning_rate": 8.7757909215956e-06,
      "loss": 0.2917,
      "step": 435
    },
    {
      "epoch": 1.74,
      "learning_rate": 8.748280605226961e-06,
      "loss": 0.2771,
      "step": 436
    },
    {
      "epoch": 1.75,
      "learning_rate": 8.720770288858322e-06,
      "loss": 0.2894,
      "step": 437
    },
    {
      "epoch": 1.75,
      "learning_rate": 8.693259972489685e-06,
      "loss": 0.315,
      "step": 438
    },
    {
      "epoch": 1.76,
      "learning_rate": 8.665749656121047e-06,
      "loss": 0.3168,
      "step": 439
    },
    {
      "epoch": 1.76,
      "learning_rate": 8.638239339752408e-06,
      "loss": 0.2843,
      "step": 440
    },
    {
      "epoch": 1.76,
      "learning_rate": 8.610729023383769e-06,
      "loss": 0.2928,
      "step": 441
    },
    {
      "epoch": 1.77,
      "learning_rate": 8.583218707015132e-06,
      "loss": 0.2945,
      "step": 442
    },
    {
      "epoch": 1.77,
      "learning_rate": 8.555708390646493e-06,
      "loss": 0.3129,
      "step": 443
    },
    {
      "epoch": 1.78,
      "learning_rate": 8.528198074277854e-06,
      "loss": 0.3279,
      "step": 444
    },
    {
      "epoch": 1.78,
      "learning_rate": 8.500687757909215e-06,
      "loss": 0.3032,
      "step": 445
    },
    {
      "epoch": 1.78,
      "learning_rate": 8.473177441540578e-06,
      "loss": 0.2923,
      "step": 446
    },
    {
      "epoch": 1.79,
      "learning_rate": 8.44566712517194e-06,
      "loss": 0.2889,
      "step": 447
    },
    {
      "epoch": 1.79,
      "learning_rate": 8.418156808803302e-06,
      "loss": 0.2625,
      "step": 448
    },
    {
      "epoch": 1.8,
      "learning_rate": 8.390646492434664e-06,
      "loss": 0.2831,
      "step": 449
    },
    {
      "epoch": 1.8,
      "learning_rate": 8.363136176066025e-06,
      "loss": 0.3488,
      "step": 450
    },
    {
      "epoch": 1.8,
      "learning_rate": 8.335625859697388e-06,
      "loss": 0.325,
      "step": 451
    },
    {
      "epoch": 1.81,
      "learning_rate": 8.308115543328749e-06,
      "loss": 0.2904,
      "step": 452
    },
    {
      "epoch": 1.81,
      "learning_rate": 8.280605226960112e-06,
      "loss": 0.315,
      "step": 453
    },
    {
      "epoch": 1.82,
      "learning_rate": 8.253094910591473e-06,
      "loss": 0.3033,
      "step": 454
    },
    {
      "epoch": 1.82,
      "learning_rate": 8.225584594222834e-06,
      "loss": 0.3404,
      "step": 455
    },
    {
      "epoch": 1.82,
      "learning_rate": 8.198074277854196e-06,
      "loss": 0.2838,
      "step": 456
    },
    {
      "epoch": 1.83,
      "learning_rate": 8.170563961485558e-06,
      "loss": 0.2823,
      "step": 457
    },
    {
      "epoch": 1.83,
      "learning_rate": 8.14305364511692e-06,
      "loss": 0.306,
      "step": 458
    },
    {
      "epoch": 1.84,
      "learning_rate": 8.115543328748281e-06,
      "loss": 0.3263,
      "step": 459
    },
    {
      "epoch": 1.84,
      "learning_rate": 8.088033012379644e-06,
      "loss": 0.3078,
      "step": 460
    },
    {
      "epoch": 1.84,
      "learning_rate": 8.060522696011005e-06,
      "loss": 0.317,
      "step": 461
    },
    {
      "epoch": 1.85,
      "learning_rate": 8.033012379642366e-06,
      "loss": 0.3439,
      "step": 462
    },
    {
      "epoch": 1.85,
      "learning_rate": 8.005502063273727e-06,
      "loss": 0.3117,
      "step": 463
    },
    {
      "epoch": 1.86,
      "learning_rate": 7.97799174690509e-06,
      "loss": 0.316,
      "step": 464
    },
    {
      "epoch": 1.86,
      "learning_rate": 7.950481430536452e-06,
      "loss": 0.328,
      "step": 465
    },
    {
      "epoch": 1.86,
      "learning_rate": 7.922971114167813e-06,
      "loss": 0.2783,
      "step": 466
    },
    {
      "epoch": 1.87,
      "learning_rate": 7.895460797799176e-06,
      "loss": 0.3354,
      "step": 467
    },
    {
      "epoch": 1.87,
      "learning_rate": 7.867950481430537e-06,
      "loss": 0.3286,
      "step": 468
    },
    {
      "epoch": 1.88,
      "learning_rate": 7.840440165061898e-06,
      "loss": 0.349,
      "step": 469
    },
    {
      "epoch": 1.88,
      "learning_rate": 7.812929848693261e-06,
      "loss": 0.3494,
      "step": 470
    },
    {
      "epoch": 1.88,
      "learning_rate": 7.785419532324622e-06,
      "loss": 0.2758,
      "step": 471
    },
    {
      "epoch": 1.89,
      "learning_rate": 7.757909215955985e-06,
      "loss": 0.3115,
      "step": 472
    },
    {
      "epoch": 1.89,
      "learning_rate": 7.730398899587346e-06,
      "loss": 0.2571,
      "step": 473
    },
    {
      "epoch": 1.9,
      "learning_rate": 7.702888583218707e-06,
      "loss": 0.3107,
      "step": 474
    },
    {
      "epoch": 1.9,
      "learning_rate": 7.67537826685007e-06,
      "loss": 0.2853,
      "step": 475
    },
    {
      "epoch": 1.9,
      "learning_rate": 7.647867950481432e-06,
      "loss": 0.3056,
      "step": 476
    },
    {
      "epoch": 1.91,
      "learning_rate": 7.620357634112793e-06,
      "loss": 0.3009,
      "step": 477
    },
    {
      "epoch": 1.91,
      "learning_rate": 7.592847317744154e-06,
      "loss": 0.298,
      "step": 478
    },
    {
      "epoch": 1.92,
      "learning_rate": 7.565337001375517e-06,
      "loss": 0.3062,
      "step": 479
    },
    {
      "epoch": 1.92,
      "learning_rate": 7.537826685006878e-06,
      "loss": 0.3205,
      "step": 480
    },
    {
      "epoch": 1.92,
      "learning_rate": 7.510316368638239e-06,
      "loss": 0.2877,
      "step": 481
    },
    {
      "epoch": 1.93,
      "learning_rate": 7.482806052269602e-06,
      "loss": 0.2873,
      "step": 482
    },
    {
      "epoch": 1.93,
      "learning_rate": 7.4552957359009634e-06,
      "loss": 0.3208,
      "step": 483
    },
    {
      "epoch": 1.94,
      "learning_rate": 7.4277854195323255e-06,
      "loss": 0.2843,
      "step": 484
    },
    {
      "epoch": 1.94,
      "learning_rate": 7.400275103163687e-06,
      "loss": 0.3534,
      "step": 485
    },
    {
      "epoch": 1.94,
      "learning_rate": 7.372764786795049e-06,
      "loss": 0.2732,
      "step": 486
    },
    {
      "epoch": 1.95,
      "learning_rate": 7.345254470426411e-06,
      "loss": 0.2605,
      "step": 487
    },
    {
      "epoch": 1.95,
      "learning_rate": 7.317744154057772e-06,
      "loss": 0.3282,
      "step": 488
    },
    {
      "epoch": 1.96,
      "learning_rate": 7.290233837689133e-06,
      "loss": 0.3356,
      "step": 489
    },
    {
      "epoch": 1.96,
      "learning_rate": 7.262723521320496e-06,
      "loss": 0.2869,
      "step": 490
    },
    {
      "epoch": 1.96,
      "learning_rate": 7.235213204951857e-06,
      "loss": 0.3214,
      "step": 491
    },
    {
      "epoch": 1.97,
      "learning_rate": 7.207702888583219e-06,
      "loss": 0.2519,
      "step": 492
    },
    {
      "epoch": 1.97,
      "learning_rate": 7.1801925722145814e-06,
      "loss": 0.3467,
      "step": 493
    },
    {
      "epoch": 1.98,
      "learning_rate": 7.152682255845943e-06,
      "loss": 0.3123,
      "step": 494
    },
    {
      "epoch": 1.98,
      "learning_rate": 7.125171939477305e-06,
      "loss": 0.3143,
      "step": 495
    },
    {
      "epoch": 1.98,
      "learning_rate": 7.097661623108666e-06,
      "loss": 0.2994,
      "step": 496
    },
    {
      "epoch": 1.99,
      "learning_rate": 7.070151306740029e-06,
      "loss": 0.3241,
      "step": 497
    },
    {
      "epoch": 1.99,
      "learning_rate": 7.04264099037139e-06,
      "loss": 0.3239,
      "step": 498
    },
    {
      "epoch": 2.0,
      "learning_rate": 7.015130674002751e-06,
      "loss": 0.2942,
      "step": 499
    },
    {
      "epoch": 2.0,
      "learning_rate": 6.9876203576341124e-06,
      "loss": 0.3155,
      "step": 500
    },
    {
      "epoch": 2.0,
      "learning_rate": 6.960110041265475e-06,
      "loss": 0.1219,
      "step": 501
    },
    {
      "epoch": 2.01,
      "learning_rate": 6.9325997248968365e-06,
      "loss": 0.0989,
      "step": 502
    },
    {
      "epoch": 2.01,
      "learning_rate": 6.905089408528199e-06,
      "loss": 0.1012,
      "step": 503
    },
    {
      "epoch": 2.02,
      "learning_rate": 6.877579092159561e-06,
      "loss": 0.0988,
      "step": 504
    },
    {
      "epoch": 2.02,
      "learning_rate": 6.850068775790923e-06,
      "loss": 0.0915,
      "step": 505
    },
    {
      "epoch": 2.02,
      "learning_rate": 6.822558459422284e-06,
      "loss": 0.097,
      "step": 506
    },
    {
      "epoch": 2.03,
      "learning_rate": 6.795048143053645e-06,
      "loss": 0.0733,
      "step": 507
    },
    {
      "epoch": 2.03,
      "learning_rate": 6.767537826685008e-06,
      "loss": 0.0898,
      "step": 508
    },
    {
      "epoch": 2.04,
      "learning_rate": 6.740027510316369e-06,
      "loss": 0.098,
      "step": 509
    },
    {
      "epoch": 2.04,
      "learning_rate": 6.7125171939477305e-06,
      "loss": 0.1005,
      "step": 510
    },
    {
      "epoch": 2.04,
      "learning_rate": 6.6850068775790925e-06,
      "loss": 0.114,
      "step": 511
    },
    {
      "epoch": 2.05,
      "learning_rate": 6.6574965612104546e-06,
      "loss": 0.0894,
      "step": 512
    },
    {
      "epoch": 2.05,
      "learning_rate": 6.629986244841816e-06,
      "loss": 0.0914,
      "step": 513
    },
    {
      "epoch": 2.06,
      "learning_rate": 6.602475928473178e-06,
      "loss": 0.0679,
      "step": 514
    },
    {
      "epoch": 2.06,
      "learning_rate": 6.57496561210454e-06,
      "loss": 0.0758,
      "step": 515
    },
    {
      "epoch": 2.06,
      "learning_rate": 6.547455295735902e-06,
      "loss": 0.0863,
      "step": 516
    },
    {
      "epoch": 2.07,
      "learning_rate": 6.519944979367263e-06,
      "loss": 0.0891,
      "step": 517
    },
    {
      "epoch": 2.07,
      "learning_rate": 6.492434662998624e-06,
      "loss": 0.0713,
      "step": 518
    },
    {
      "epoch": 2.08,
      "learning_rate": 6.464924346629987e-06,
      "loss": 0.0867,
      "step": 519
    },
    {
      "epoch": 2.08,
      "learning_rate": 6.4374140302613485e-06,
      "loss": 0.0871,
      "step": 520
    },
    {
      "epoch": 2.08,
      "learning_rate": 6.40990371389271e-06,
      "loss": 0.0778,
      "step": 521
    },
    {
      "epoch": 2.09,
      "learning_rate": 6.382393397524072e-06,
      "loss": 0.0896,
      "step": 522
    },
    {
      "epoch": 2.09,
      "learning_rate": 6.354883081155434e-06,
      "loss": 0.0913,
      "step": 523
    },
    {
      "epoch": 2.1,
      "learning_rate": 6.327372764786796e-06,
      "loss": 0.0807,
      "step": 524
    },
    {
      "epoch": 2.1,
      "learning_rate": 6.299862448418157e-06,
      "loss": 0.0811,
      "step": 525
    },
    {
      "epoch": 2.1,
      "learning_rate": 6.27235213204952e-06,
      "loss": 0.0919,
      "step": 526
    },
    {
      "epoch": 2.11,
      "learning_rate": 6.244841815680881e-06,
      "loss": 0.0605,
      "step": 527
    },
    {
      "epoch": 2.11,
      "learning_rate": 6.217331499312242e-06,
      "loss": 0.0743,
      "step": 528
    },
    {
      "epoch": 2.12,
      "learning_rate": 6.189821182943604e-06,
      "loss": 0.0813,
      "step": 529
    },
    {
      "epoch": 2.12,
      "learning_rate": 6.1623108665749665e-06,
      "loss": 0.0786,
      "step": 530
    },
    {
      "epoch": 2.12,
      "learning_rate": 6.134800550206328e-06,
      "loss": 0.0809,
      "step": 531
    },
    {
      "epoch": 2.13,
      "learning_rate": 6.10729023383769e-06,
      "loss": 0.0692,
      "step": 532
    },
    {
      "epoch": 2.13,
      "learning_rate": 6.079779917469052e-06,
      "loss": 0.0696,
      "step": 533
    },
    {
      "epoch": 2.14,
      "learning_rate": 6.052269601100413e-06,
      "loss": 0.0947,
      "step": 534
    },
    {
      "epoch": 2.14,
      "learning_rate": 6.024759284731775e-06,
      "loss": 0.0876,
      "step": 535
    },
    {
      "epoch": 2.14,
      "learning_rate": 5.997248968363136e-06,
      "loss": 0.0816,
      "step": 536
    },
    {
      "epoch": 2.15,
      "learning_rate": 5.969738651994499e-06,
      "loss": 0.0722,
      "step": 537
    },
    {
      "epoch": 2.15,
      "learning_rate": 5.94222833562586e-06,
      "loss": 0.0823,
      "step": 538
    },
    {
      "epoch": 2.16,
      "learning_rate": 5.914718019257222e-06,
      "loss": 0.0916,
      "step": 539
    },
    {
      "epoch": 2.16,
      "learning_rate": 5.887207702888584e-06,
      "loss": 0.0696,
      "step": 540
    },
    {
      "epoch": 2.16,
      "learning_rate": 5.859697386519946e-06,
      "loss": 0.107,
      "step": 541
    },
    {
      "epoch": 2.17,
      "learning_rate": 5.832187070151307e-06,
      "loss": 0.0832,
      "step": 542
    },
    {
      "epoch": 2.17,
      "learning_rate": 5.804676753782669e-06,
      "loss": 0.0955,
      "step": 543
    },
    {
      "epoch": 2.18,
      "learning_rate": 5.777166437414031e-06,
      "loss": 0.0934,
      "step": 544
    },
    {
      "epoch": 2.18,
      "learning_rate": 5.749656121045393e-06,
      "loss": 0.0974,
      "step": 545
    },
    {
      "epoch": 2.18,
      "learning_rate": 5.722145804676754e-06,
      "loss": 0.076,
      "step": 546
    },
    {
      "epoch": 2.19,
      "learning_rate": 5.6946354883081155e-06,
      "loss": 0.0603,
      "step": 547
    },
    {
      "epoch": 2.19,
      "learning_rate": 5.667125171939478e-06,
      "loss": 0.1073,
      "step": 548
    },
    {
      "epoch": 2.2,
      "learning_rate": 5.63961485557084e-06,
      "loss": 0.0638,
      "step": 549
    },
    {
      "epoch": 2.2,
      "learning_rate": 5.612104539202201e-06,
      "loss": 0.0747,
      "step": 550
    },
    {
      "epoch": 2.2,
      "learning_rate": 5.584594222833563e-06,
      "loss": 0.0803,
      "step": 551
    },
    {
      "epoch": 2.21,
      "learning_rate": 5.557083906464925e-06,
      "loss": 0.0687,
      "step": 552
    },
    {
      "epoch": 2.21,
      "learning_rate": 5.529573590096287e-06,
      "loss": 0.0822,
      "step": 553
    },
    {
      "epoch": 2.22,
      "learning_rate": 5.502063273727648e-06,
      "loss": 0.087,
      "step": 554
    },
    {
      "epoch": 2.22,
      "learning_rate": 5.47455295735901e-06,
      "loss": 0.0738,
      "step": 555
    },
    {
      "epoch": 2.22,
      "learning_rate": 5.447042640990372e-06,
      "loss": 0.072,
      "step": 556
    },
    {
      "epoch": 2.23,
      "learning_rate": 5.4195323246217335e-06,
      "loss": 0.08,
      "step": 557
    },
    {
      "epoch": 2.23,
      "learning_rate": 5.392022008253095e-06,
      "loss": 0.1007,
      "step": 558
    },
    {
      "epoch": 2.24,
      "learning_rate": 5.364511691884458e-06,
      "loss": 0.0886,
      "step": 559
    },
    {
      "epoch": 2.24,
      "learning_rate": 5.337001375515819e-06,
      "loss": 0.0782,
      "step": 560
    },
    {
      "epoch": 2.24,
      "learning_rate": 5.30949105914718e-06,
      "loss": 0.0879,
      "step": 561
    },
    {
      "epoch": 2.25,
      "learning_rate": 5.281980742778542e-06,
      "loss": 0.0709,
      "step": 562
    },
    {
      "epoch": 2.25,
      "learning_rate": 5.254470426409904e-06,
      "loss": 0.0539,
      "step": 563
    },
    {
      "epoch": 2.26,
      "learning_rate": 5.226960110041266e-06,
      "loss": 0.0815,
      "step": 564
    },
    {
      "epoch": 2.26,
      "learning_rate": 5.199449793672627e-06,
      "loss": 0.0745,
      "step": 565
    },
    {
      "epoch": 2.26,
      "learning_rate": 5.17193947730399e-06,
      "loss": 0.0769,
      "step": 566
    },
    {
      "epoch": 2.27,
      "learning_rate": 5.1444291609353515e-06,
      "loss": 0.0723,
      "step": 567
    },
    {
      "epoch": 2.27,
      "learning_rate": 5.116918844566713e-06,
      "loss": 0.0713,
      "step": 568
    },
    {
      "epoch": 2.28,
      "learning_rate": 5.089408528198074e-06,
      "loss": 0.085,
      "step": 569
    },
    {
      "epoch": 2.28,
      "learning_rate": 5.061898211829437e-06,
      "loss": 0.0857,
      "step": 570
    },
    {
      "epoch": 2.28,
      "learning_rate": 5.034387895460798e-06,
      "loss": 0.0728,
      "step": 571
    },
    {
      "epoch": 2.29,
      "learning_rate": 5.00687757909216e-06,
      "loss": 0.0733,
      "step": 572
    },
    {
      "epoch": 2.29,
      "learning_rate": 4.979367262723522e-06,
      "loss": 0.0727,
      "step": 573
    },
    {
      "epoch": 2.3,
      "learning_rate": 4.951856946354883e-06,
      "loss": 0.0765,
      "step": 574
    },
    {
      "epoch": 2.3,
      "learning_rate": 4.9243466299862454e-06,
      "loss": 0.0732,
      "step": 575
    },
    {
      "epoch": 2.3,
      "learning_rate": 4.8968363136176075e-06,
      "loss": 0.0889,
      "step": 576
    },
    {
      "epoch": 2.31,
      "learning_rate": 4.869325997248969e-06,
      "loss": 0.0803,
      "step": 577
    },
    {
      "epoch": 2.31,
      "learning_rate": 4.841815680880331e-06,
      "loss": 0.0759,
      "step": 578
    },
    {
      "epoch": 2.32,
      "learning_rate": 4.814305364511692e-06,
      "loss": 0.0818,
      "step": 579
    },
    {
      "epoch": 2.32,
      "learning_rate": 4.786795048143054e-06,
      "loss": 0.0996,
      "step": 580
    },
    {
      "epoch": 2.32,
      "learning_rate": 4.759284731774415e-06,
      "loss": 0.0707,
      "step": 581
    },
    {
      "epoch": 2.33,
      "learning_rate": 4.731774415405777e-06,
      "loss": 0.0908,
      "step": 582
    },
    {
      "epoch": 2.33,
      "learning_rate": 4.704264099037139e-06,
      "loss": 0.0724,
      "step": 583
    },
    {
      "epoch": 2.34,
      "learning_rate": 4.676753782668501e-06,
      "loss": 0.072,
      "step": 584
    },
    {
      "epoch": 2.34,
      "learning_rate": 4.649243466299863e-06,
      "loss": 0.071,
      "step": 585
    },
    {
      "epoch": 2.34,
      "learning_rate": 4.621733149931225e-06,
      "loss": 0.0762,
      "step": 586
    },
    {
      "epoch": 2.35,
      "learning_rate": 4.594222833562587e-06,
      "loss": 0.0753,
      "step": 587
    },
    {
      "epoch": 2.35,
      "learning_rate": 4.566712517193948e-06,
      "loss": 0.0938,
      "step": 588
    },
    {
      "epoch": 2.36,
      "learning_rate": 4.53920220082531e-06,
      "loss": 0.0718,
      "step": 589
    },
    {
      "epoch": 2.36,
      "learning_rate": 4.511691884456671e-06,
      "loss": 0.0851,
      "step": 590
    },
    {
      "epoch": 2.36,
      "learning_rate": 4.484181568088033e-06,
      "loss": 0.0891,
      "step": 591
    },
    {
      "epoch": 2.37,
      "learning_rate": 4.456671251719395e-06,
      "loss": 0.0827,
      "step": 592
    },
    {
      "epoch": 2.37,
      "learning_rate": 4.429160935350757e-06,
      "loss": 0.0808,
      "step": 593
    },
    {
      "epoch": 2.38,
      "learning_rate": 4.4016506189821186e-06,
      "loss": 0.0902,
      "step": 594
    },
    {
      "epoch": 2.38,
      "learning_rate": 4.374140302613481e-06,
      "loss": 0.0676,
      "step": 595
    },
    {
      "epoch": 2.38,
      "learning_rate": 4.346629986244843e-06,
      "loss": 0.0716,
      "step": 596
    },
    {
      "epoch": 2.39,
      "learning_rate": 4.319119669876204e-06,
      "loss": 0.0769,
      "step": 597
    },
    {
      "epoch": 2.39,
      "learning_rate": 4.291609353507566e-06,
      "loss": 0.0815,
      "step": 598
    },
    {
      "epoch": 2.4,
      "learning_rate": 4.264099037138927e-06,
      "loss": 0.0698,
      "step": 599
    },
    {
      "epoch": 2.4,
      "learning_rate": 4.236588720770289e-06,
      "loss": 0.0777,
      "step": 600
    },
    {
      "epoch": 2.4,
      "learning_rate": 4.209078404401651e-06,
      "loss": 0.0933,
      "step": 601
    },
    {
      "epoch": 2.41,
      "learning_rate": 4.1815680880330125e-06,
      "loss": 0.0835,
      "step": 602
    },
    {
      "epoch": 2.41,
      "learning_rate": 4.1540577716643745e-06,
      "loss": 0.0704,
      "step": 603
    },
    {
      "epoch": 2.42,
      "learning_rate": 4.1265474552957366e-06,
      "loss": 0.0784,
      "step": 604
    },
    {
      "epoch": 2.42,
      "learning_rate": 4.099037138927098e-06,
      "loss": 0.0867,
      "step": 605
    },
    {
      "epoch": 2.42,
      "learning_rate": 4.07152682255846e-06,
      "loss": 0.0947,
      "step": 606
    },
    {
      "epoch": 2.43,
      "learning_rate": 4.044016506189822e-06,
      "loss": 0.0907,
      "step": 607
    },
    {
      "epoch": 2.43,
      "learning_rate": 4.016506189821183e-06,
      "loss": 0.0735,
      "step": 608
    },
    {
      "epoch": 2.44,
      "learning_rate": 3.988995873452545e-06,
      "loss": 0.0667,
      "step": 609
    },
    {
      "epoch": 2.44,
      "learning_rate": 3.961485557083906e-06,
      "loss": 0.0788,
      "step": 610
    },
    {
      "epoch": 2.44,
      "learning_rate": 3.933975240715268e-06,
      "loss": 0.0684,
      "step": 611
    },
    {
      "epoch": 2.45,
      "learning_rate": 3.9064649243466305e-06,
      "loss": 0.078,
      "step": 612
    },
    {
      "epoch": 2.45,
      "learning_rate": 3.8789546079779925e-06,
      "loss": 0.103,
      "step": 613
    },
    {
      "epoch": 2.46,
      "learning_rate": 3.851444291609354e-06,
      "loss": 0.0702,
      "step": 614
    },
    {
      "epoch": 2.46,
      "learning_rate": 3.823933975240716e-06,
      "loss": 0.0681,
      "step": 615
    },
    {
      "epoch": 2.46,
      "learning_rate": 3.796423658872077e-06,
      "loss": 0.0629,
      "step": 616
    },
    {
      "epoch": 2.47,
      "learning_rate": 3.768913342503439e-06,
      "loss": 0.0641,
      "step": 617
    },
    {
      "epoch": 2.47,
      "learning_rate": 3.741403026134801e-06,
      "loss": 0.069,
      "step": 618
    },
    {
      "epoch": 2.48,
      "learning_rate": 3.7138927097661627e-06,
      "loss": 0.0894,
      "step": 619
    },
    {
      "epoch": 2.48,
      "learning_rate": 3.6863823933975244e-06,
      "loss": 0.0581,
      "step": 620
    },
    {
      "epoch": 2.48,
      "learning_rate": 3.658872077028886e-06,
      "loss": 0.076,
      "step": 621
    },
    {
      "epoch": 2.49,
      "learning_rate": 3.631361760660248e-06,
      "loss": 0.079,
      "step": 622
    },
    {
      "epoch": 2.49,
      "learning_rate": 3.6038514442916097e-06,
      "loss": 0.0614,
      "step": 623
    },
    {
      "epoch": 2.5,
      "learning_rate": 3.5763411279229713e-06,
      "loss": 0.0906,
      "step": 624
    },
    {
      "epoch": 2.5,
      "learning_rate": 3.548830811554333e-06,
      "loss": 0.0733,
      "step": 625
    },
    {
      "epoch": 2.5,
      "learning_rate": 3.521320495185695e-06,
      "loss": 0.0835,
      "step": 626
    },
    {
      "epoch": 2.51,
      "learning_rate": 3.4938101788170562e-06,
      "loss": 0.0871,
      "step": 627
    },
    {
      "epoch": 2.51,
      "learning_rate": 3.4662998624484183e-06,
      "loss": 0.0605,
      "step": 628
    },
    {
      "epoch": 2.52,
      "learning_rate": 3.4387895460797803e-06,
      "loss": 0.0884,
      "step": 629
    },
    {
      "epoch": 2.52,
      "learning_rate": 3.411279229711142e-06,
      "loss": 0.0839,
      "step": 630
    },
    {
      "epoch": 2.52,
      "learning_rate": 3.383768913342504e-06,
      "loss": 0.0678,
      "step": 631
    },
    {
      "epoch": 2.53,
      "learning_rate": 3.3562585969738652e-06,
      "loss": 0.086,
      "step": 632
    },
    {
      "epoch": 2.53,
      "learning_rate": 3.3287482806052273e-06,
      "loss": 0.0747,
      "step": 633
    },
    {
      "epoch": 2.54,
      "learning_rate": 3.301237964236589e-06,
      "loss": 0.081,
      "step": 634
    },
    {
      "epoch": 2.54,
      "learning_rate": 3.273727647867951e-06,
      "loss": 0.0888,
      "step": 635
    },
    {
      "epoch": 2.54,
      "learning_rate": 3.246217331499312e-06,
      "loss": 0.0762,
      "step": 636
    },
    {
      "epoch": 2.55,
      "learning_rate": 3.2187070151306742e-06,
      "loss": 0.07,
      "step": 637
    },
    {
      "epoch": 2.55,
      "learning_rate": 3.191196698762036e-06,
      "loss": 0.0863,
      "step": 638
    },
    {
      "epoch": 2.56,
      "learning_rate": 3.163686382393398e-06,
      "loss": 0.087,
      "step": 639
    },
    {
      "epoch": 2.56,
      "learning_rate": 3.13617606602476e-06,
      "loss": 0.0826,
      "step": 640
    },
    {
      "epoch": 2.56,
      "learning_rate": 3.108665749656121e-06,
      "loss": 0.084,
      "step": 641
    },
    {
      "epoch": 2.57,
      "learning_rate": 3.0811554332874832e-06,
      "loss": 0.0635,
      "step": 642
    },
    {
      "epoch": 2.57,
      "learning_rate": 3.053645116918845e-06,
      "loss": 0.0718,
      "step": 643
    },
    {
      "epoch": 2.58,
      "learning_rate": 3.0261348005502065e-06,
      "loss": 0.0676,
      "step": 644
    },
    {
      "epoch": 2.58,
      "learning_rate": 2.998624484181568e-06,
      "loss": 0.0934,
      "step": 645
    },
    {
      "epoch": 2.58,
      "learning_rate": 2.97111416781293e-06,
      "loss": 0.0605,
      "step": 646
    },
    {
      "epoch": 2.59,
      "learning_rate": 2.943603851444292e-06,
      "loss": 0.0711,
      "step": 647
    },
    {
      "epoch": 2.59,
      "learning_rate": 2.9160935350756535e-06,
      "loss": 0.0872,
      "step": 648
    },
    {
      "epoch": 2.6,
      "learning_rate": 2.8885832187070155e-06,
      "loss": 0.0681,
      "step": 649
    },
    {
      "epoch": 2.6,
      "learning_rate": 2.861072902338377e-06,
      "loss": 0.0856,
      "step": 650
    },
    {
      "epoch": 2.6,
      "learning_rate": 2.833562585969739e-06,
      "loss": 0.0611,
      "step": 651
    },
    {
      "epoch": 2.61,
      "learning_rate": 2.8060522696011004e-06,
      "loss": 0.0888,
      "step": 652
    },
    {
      "epoch": 2.61,
      "learning_rate": 2.7785419532324625e-06,
      "loss": 0.0741,
      "step": 653
    },
    {
      "epoch": 2.62,
      "learning_rate": 2.751031636863824e-06,
      "loss": 0.0763,
      "step": 654
    },
    {
      "epoch": 2.62,
      "learning_rate": 2.723521320495186e-06,
      "loss": 0.0909,
      "step": 655
    },
    {
      "epoch": 2.62,
      "learning_rate": 2.6960110041265474e-06,
      "loss": 0.0681,
      "step": 656
    },
    {
      "epoch": 2.63,
      "learning_rate": 2.6685006877579094e-06,
      "loss": 0.079,
      "step": 657
    },
    {
      "epoch": 2.63,
      "learning_rate": 2.640990371389271e-06,
      "loss": 0.078,
      "step": 658
    },
    {
      "epoch": 2.64,
      "learning_rate": 2.613480055020633e-06,
      "loss": 0.0895,
      "step": 659
    },
    {
      "epoch": 2.64,
      "learning_rate": 2.585969738651995e-06,
      "loss": 0.0763,
      "step": 660
    },
    {
      "epoch": 2.64,
      "learning_rate": 2.5584594222833564e-06,
      "loss": 0.0683,
      "step": 661
    },
    {
      "epoch": 2.65,
      "learning_rate": 2.5309491059147184e-06,
      "loss": 0.0771,
      "step": 662
    },
    {
      "epoch": 2.65,
      "learning_rate": 2.50343878954608e-06,
      "loss": 0.0685,
      "step": 663
    },
    {
      "epoch": 2.66,
      "learning_rate": 2.4759284731774417e-06,
      "loss": 0.0835,
      "step": 664
    },
    {
      "epoch": 2.66,
      "learning_rate": 2.4484181568088037e-06,
      "loss": 0.098,
      "step": 665
    },
    {
      "epoch": 2.66,
      "learning_rate": 2.4209078404401654e-06,
      "loss": 0.0889,
      "step": 666
    },
    {
      "epoch": 2.67,
      "learning_rate": 2.393397524071527e-06,
      "loss": 0.0776,
      "step": 667
    },
    {
      "epoch": 2.67,
      "learning_rate": 2.3658872077028886e-06,
      "loss": 0.0817,
      "step": 668
    },
    {
      "epoch": 2.68,
      "learning_rate": 2.3383768913342507e-06,
      "loss": 0.0595,
      "step": 669
    },
    {
      "epoch": 2.68,
      "learning_rate": 2.3108665749656123e-06,
      "loss": 0.075,
      "step": 670
    },
    {
      "epoch": 2.68,
      "learning_rate": 2.283356258596974e-06,
      "loss": 0.0643,
      "step": 671
    },
    {
      "epoch": 2.69,
      "learning_rate": 2.2558459422283356e-06,
      "loss": 0.078,
      "step": 672
    },
    {
      "epoch": 2.69,
      "learning_rate": 2.2283356258596976e-06,
      "loss": 0.0559,
      "step": 673
    },
    {
      "epoch": 2.7,
      "learning_rate": 2.2008253094910593e-06,
      "loss": 0.0719,
      "step": 674
    },
    {
      "epoch": 2.7,
      "learning_rate": 2.1733149931224213e-06,
      "loss": 0.0862,
      "step": 675
    },
    {
      "epoch": 2.7,
      "learning_rate": 2.145804676753783e-06,
      "loss": 0.0607,
      "step": 676
    },
    {
      "epoch": 2.71,
      "learning_rate": 2.1182943603851446e-06,
      "loss": 0.0696,
      "step": 677
    },
    {
      "epoch": 2.71,
      "learning_rate": 2.0907840440165062e-06,
      "loss": 0.0795,
      "step": 678
    },
    {
      "epoch": 2.72,
      "learning_rate": 2.0632737276478683e-06,
      "loss": 0.0901,
      "step": 679
    },
    {
      "epoch": 2.72,
      "learning_rate": 2.03576341127923e-06,
      "loss": 0.0813,
      "step": 680
    },
    {
      "epoch": 2.72,
      "learning_rate": 2.0082530949105915e-06,
      "loss": 0.0811,
      "step": 681
    },
    {
      "epoch": 2.73,
      "learning_rate": 1.980742778541953e-06,
      "loss": 0.0665,
      "step": 682
    },
    {
      "epoch": 2.73,
      "learning_rate": 1.9532324621733152e-06,
      "loss": 0.0742,
      "step": 683
    },
    {
      "epoch": 2.74,
      "learning_rate": 1.925722145804677e-06,
      "loss": 0.0786,
      "step": 684
    },
    {
      "epoch": 2.74,
      "learning_rate": 1.8982118294360385e-06,
      "loss": 0.0844,
      "step": 685
    },
    {
      "epoch": 2.74,
      "learning_rate": 1.8707015130674006e-06,
      "loss": 0.0747,
      "step": 686
    },
    {
      "epoch": 2.75,
      "learning_rate": 1.8431911966987622e-06,
      "loss": 0.0818,
      "step": 687
    },
    {
      "epoch": 2.75,
      "learning_rate": 1.815680880330124e-06,
      "loss": 0.0938,
      "step": 688
    },
    {
      "epoch": 2.76,
      "learning_rate": 1.7881705639614857e-06,
      "loss": 0.0863,
      "step": 689
    },
    {
      "epoch": 2.76,
      "learning_rate": 1.7606602475928475e-06,
      "loss": 0.0875,
      "step": 690
    },
    {
      "epoch": 2.76,
      "learning_rate": 1.7331499312242091e-06,
      "loss": 0.0824,
      "step": 691
    },
    {
      "epoch": 2.77,
      "learning_rate": 1.705639614855571e-06,
      "loss": 0.0751,
      "step": 692
    },
    {
      "epoch": 2.77,
      "learning_rate": 1.6781292984869326e-06,
      "loss": 0.08,
      "step": 693
    },
    {
      "epoch": 2.78,
      "learning_rate": 1.6506189821182945e-06,
      "loss": 0.0569,
      "step": 694
    },
    {
      "epoch": 2.78,
      "learning_rate": 1.623108665749656e-06,
      "loss": 0.0696,
      "step": 695
    },
    {
      "epoch": 2.78,
      "learning_rate": 1.595598349381018e-06,
      "loss": 0.0717,
      "step": 696
    },
    {
      "epoch": 2.79,
      "learning_rate": 1.56808803301238e-06,
      "loss": 0.0548,
      "step": 697
    },
    {
      "epoch": 2.79,
      "learning_rate": 1.5405777166437416e-06,
      "loss": 0.084,
      "step": 698
    },
    {
      "epoch": 2.8,
      "learning_rate": 1.5130674002751033e-06,
      "loss": 0.085,
      "step": 699
    },
    {
      "epoch": 2.8,
      "learning_rate": 1.485557083906465e-06,
      "loss": 0.0848,
      "step": 700
    },
    {
      "epoch": 2.8,
      "learning_rate": 1.4580467675378267e-06,
      "loss": 0.0793,
      "step": 701
    },
    {
      "epoch": 2.81,
      "learning_rate": 1.4305364511691886e-06,
      "loss": 0.0653,
      "step": 702
    },
    {
      "epoch": 2.81,
      "learning_rate": 1.4030261348005502e-06,
      "loss": 0.0813,
      "step": 703
    },
    {
      "epoch": 2.82,
      "learning_rate": 1.375515818431912e-06,
      "loss": 0.0653,
      "step": 704
    },
    {
      "epoch": 2.82,
      "learning_rate": 1.3480055020632737e-06,
      "loss": 0.0773,
      "step": 705
    },
    {
      "epoch": 2.82,
      "learning_rate": 1.3204951856946355e-06,
      "loss": 0.093,
      "step": 706
    },
    {
      "epoch": 2.83,
      "learning_rate": 1.2929848693259976e-06,
      "loss": 0.0606,
      "step": 707
    },
    {
      "epoch": 2.83,
      "learning_rate": 1.2654745529573592e-06,
      "loss": 0.0667,
      "step": 708
    },
    {
      "epoch": 2.84,
      "learning_rate": 1.2379642365887208e-06,
      "loss": 0.0685,
      "step": 709
    },
    {
      "epoch": 2.84,
      "learning_rate": 1.2104539202200827e-06,
      "loss": 0.0701,
      "step": 710
    },
    {
      "epoch": 2.84,
      "learning_rate": 1.1829436038514443e-06,
      "loss": 0.0699,
      "step": 711
    },
    {
      "epoch": 2.85,
      "learning_rate": 1.1554332874828062e-06,
      "loss": 0.0796,
      "step": 712
    },
    {
      "epoch": 2.85,
      "learning_rate": 1.1279229711141678e-06,
      "loss": 0.0907,
      "step": 713
    },
    {
      "epoch": 2.86,
      "learning_rate": 1.1004126547455296e-06,
      "loss": 0.0626,
      "step": 714
    },
    {
      "epoch": 2.86,
      "learning_rate": 1.0729023383768915e-06,
      "loss": 0.0551,
      "step": 715
    },
    {
      "epoch": 2.86,
      "learning_rate": 1.0453920220082531e-06,
      "loss": 0.069,
      "step": 716
    },
    {
      "epoch": 2.87,
      "learning_rate": 1.017881705639615e-06,
      "loss": 0.0748,
      "step": 717
    },
    {
      "epoch": 2.87,
      "learning_rate": 9.903713892709766e-07,
      "loss": 0.0696,
      "step": 718
    },
    {
      "epoch": 2.88,
      "learning_rate": 9.628610729023384e-07,
      "loss": 0.064,
      "step": 719
    },
    {
      "epoch": 2.88,
      "learning_rate": 9.353507565337003e-07,
      "loss": 0.0614,
      "step": 720
    },
    {
      "epoch": 2.88,
      "learning_rate": 9.07840440165062e-07,
      "loss": 0.0667,
      "step": 721
    },
    {
      "epoch": 2.89,
      "learning_rate": 8.803301237964238e-07,
      "loss": 0.0737,
      "step": 722
    },
    {
      "epoch": 2.89,
      "learning_rate": 8.528198074277855e-07,
      "loss": 0.0585,
      "step": 723
    },
    {
      "epoch": 2.9,
      "learning_rate": 8.253094910591472e-07,
      "loss": 0.0814,
      "step": 724
    },
    {
      "epoch": 2.9,
      "learning_rate": 7.97799174690509e-07,
      "loss": 0.0751,
      "step": 725
    },
    {
      "epoch": 2.9,
      "learning_rate": 7.702888583218708e-07,
      "loss": 0.0693,
      "step": 726
    },
    {
      "epoch": 2.91,
      "learning_rate": 7.427785419532325e-07,
      "loss": 0.0771,
      "step": 727
    },
    {
      "epoch": 2.91,
      "learning_rate": 7.152682255845943e-07,
      "loss": 0.0678,
      "step": 728
    },
    {
      "epoch": 2.92,
      "learning_rate": 6.87757909215956e-07,
      "loss": 0.0857,
      "step": 729
    },
    {
      "epoch": 2.92,
      "learning_rate": 6.602475928473178e-07,
      "loss": 0.0857,
      "step": 730
    },
    {
      "epoch": 2.92,
      "learning_rate": 6.327372764786796e-07,
      "loss": 0.0864,
      "step": 731
    },
    {
      "epoch": 2.93,
      "learning_rate": 6.052269601100413e-07,
      "loss": 0.0851,
      "step": 732
    },
    {
      "epoch": 2.93,
      "learning_rate": 5.777166437414031e-07,
      "loss": 0.0889,
      "step": 733
    },
    {
      "epoch": 2.94,
      "learning_rate": 5.502063273727648e-07,
      "loss": 0.0771,
      "step": 734
    },
    {
      "epoch": 2.94,
      "learning_rate": 5.226960110041266e-07,
      "loss": 0.0822,
      "step": 735
    },
    {
      "epoch": 2.94,
      "learning_rate": 4.951856946354883e-07,
      "loss": 0.0642,
      "step": 736
    },
    {
      "epoch": 2.95,
      "learning_rate": 4.6767537826685014e-07,
      "loss": 0.0765,
      "step": 737
    },
    {
      "epoch": 2.95,
      "learning_rate": 4.401650618982119e-07,
      "loss": 0.0821,
      "step": 738
    },
    {
      "epoch": 2.96,
      "learning_rate": 4.126547455295736e-07,
      "loss": 0.0821,
      "step": 739
    },
    {
      "epoch": 2.96,
      "learning_rate": 3.851444291609354e-07,
      "loss": 0.0783,
      "step": 740
    },
    {
      "epoch": 2.96,
      "learning_rate": 3.5763411279229714e-07,
      "loss": 0.0722,
      "step": 741
    },
    {
      "epoch": 2.97,
      "learning_rate": 3.301237964236589e-07,
      "loss": 0.0935,
      "step": 742
    },
    {
      "epoch": 2.97,
      "learning_rate": 3.0261348005502067e-07,
      "loss": 0.0929,
      "step": 743
    },
    {
      "epoch": 2.98,
      "learning_rate": 2.751031636863824e-07,
      "loss": 0.0869,
      "step": 744
    },
    {
      "epoch": 2.98,
      "learning_rate": 2.4759284731774415e-07,
      "loss": 0.1,
      "step": 745
    },
    {
      "epoch": 2.98,
      "learning_rate": 2.2008253094910594e-07,
      "loss": 0.0941,
      "step": 746
    },
    {
      "epoch": 2.99,
      "learning_rate": 1.925722145804677e-07,
      "loss": 0.0596,
      "step": 747
    },
    {
      "epoch": 2.99,
      "learning_rate": 1.6506189821182944e-07,
      "loss": 0.0703,
      "step": 748
    },
    {
      "epoch": 3.0,
      "learning_rate": 1.375515818431912e-07,
      "loss": 0.0806,
      "step": 749
    },
    {
      "epoch": 3.0,
      "learning_rate": 1.1004126547455297e-07,
      "loss": 0.0651,
      "step": 750
    },
    {
      "epoch": 3.0,
      "step": 750,
      "total_flos": 21264552591360.0,
      "train_loss": 0.3743656844943762,
      "train_runtime": 11966.0892,
      "train_samples_per_second": 8.023,
      "train_steps_per_second": 0.063
    }
  ],
  "max_steps": 750,
  "num_train_epochs": 3,
  "total_flos": 21264552591360.0,
  "trial_name": null,
  "trial_params": null
}