{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 6.329113924050633,
  "global_step": 500,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.01,
      "learning_rate": 5e-05,
      "loss": 1.0641,
      "step": 1
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.0001,
      "loss": 0.9056,
      "step": 2
    },
    {
      "epoch": 0.04,
      "learning_rate": 9.999975227016531e-05,
      "loss": 0.9593,
      "step": 3
    },
    {
      "epoch": 0.05,
      "learning_rate": 9.999900908311602e-05,
      "loss": 0.973,
      "step": 4
    },
    {
      "epoch": 0.06,
      "learning_rate": 9.999777044621652e-05,
      "loss": 1.0652,
      "step": 5
    },
    {
      "epoch": 0.08,
      "learning_rate": 9.999603637174071e-05,
      "loss": 0.9212,
      "step": 6
    },
    {
      "epoch": 0.09,
      "learning_rate": 9.999380687687188e-05,
      "loss": 0.899,
      "step": 7
    },
    {
      "epoch": 0.1,
      "learning_rate": 9.999108198370249e-05,
      "loss": 0.7079,
      "step": 8
    },
    {
      "epoch": 0.11,
      "learning_rate": 9.998786171923407e-05,
      "loss": 0.7139,
      "step": 9
    },
    {
      "epoch": 0.13,
      "learning_rate": 9.998414611537681e-05,
      "loss": 0.8258,
      "step": 10
    },
    {
      "epoch": 0.14,
      "learning_rate": 9.997993520894937e-05,
      "loss": 0.7415,
      "step": 11
    },
    {
      "epoch": 0.15,
      "learning_rate": 9.997522904167844e-05,
      "loss": 0.8233,
      "step": 12
    },
    {
      "epoch": 0.16,
      "learning_rate": 9.997002766019832e-05,
      "loss": 0.8698,
      "step": 13
    },
    {
      "epoch": 0.18,
      "learning_rate": 9.996433111605052e-05,
      "loss": 0.6354,
      "step": 14
    },
    {
      "epoch": 0.19,
      "learning_rate": 9.99581394656832e-05,
      "loss": 0.6136,
      "step": 15
    },
    {
      "epoch": 0.2,
      "learning_rate": 9.995145277045061e-05,
      "loss": 0.7363,
      "step": 16
    },
    {
      "epoch": 0.22,
      "learning_rate": 9.994427109661253e-05,
      "loss": 0.6065,
      "step": 17
    },
    {
      "epoch": 0.23,
      "learning_rate": 9.993659451533353e-05,
      "loss": 0.6152,
      "step": 18
    },
    {
      "epoch": 0.24,
      "learning_rate": 9.992842310268233e-05,
      "loss": 0.7021,
      "step": 19
    },
    {
      "epoch": 0.25,
      "learning_rate": 9.991975693963107e-05,
      "loss": 0.5468,
      "step": 20
    },
    {
      "epoch": 0.27,
      "learning_rate": 9.99105961120544e-05,
      "loss": 0.5437,
      "step": 21
    },
    {
      "epoch": 0.28,
      "learning_rate": 9.990094071072877e-05,
      "loss": 0.6458,
      "step": 22
    },
    {
      "epoch": 0.29,
      "learning_rate": 9.989079083133139e-05,
      "loss": 0.6744,
      "step": 23
    },
    {
      "epoch": 0.3,
      "learning_rate": 9.988014657443941e-05,
      "loss": 0.5592,
      "step": 24
    },
    {
      "epoch": 0.32,
      "learning_rate": 9.986900804552878e-05,
      "loss": 0.5351,
      "step": 25
    },
    {
      "epoch": 0.33,
      "learning_rate": 9.985737535497337e-05,
      "loss": 0.5705,
      "step": 26
    },
    {
      "epoch": 0.34,
      "learning_rate": 9.984524861804376e-05,
      "loss": 0.5696,
      "step": 27
    },
    {
      "epoch": 0.35,
      "learning_rate": 9.983262795490613e-05,
      "loss": 0.588,
      "step": 28
    },
    {
      "epoch": 0.37,
      "learning_rate": 9.981951349062106e-05,
      "loss": 0.627,
      "step": 29
    },
    {
      "epoch": 0.38,
      "learning_rate": 9.980590535514233e-05,
      "loss": 0.5702,
      "step": 30
    },
    {
      "epoch": 0.39,
      "learning_rate": 9.979180368331558e-05,
      "loss": 0.5907,
      "step": 31
    },
    {
      "epoch": 0.41,
      "learning_rate": 9.9777208614877e-05,
      "loss": 0.4817,
      "step": 32
    },
    {
      "epoch": 0.42,
      "learning_rate": 9.976212029445194e-05,
      "loss": 0.5789,
      "step": 33
    },
    {
      "epoch": 0.43,
      "learning_rate": 9.97465388715535e-05,
      "loss": 0.5108,
      "step": 34
    },
    {
      "epoch": 0.44,
      "learning_rate": 9.9730464500581e-05,
      "loss": 0.557,
      "step": 35
    },
    {
      "epoch": 0.46,
      "learning_rate": 9.971389734081848e-05,
      "loss": 0.5771,
      "step": 36
    },
    {
      "epoch": 0.47,
      "learning_rate": 9.969683755643317e-05,
      "loss": 0.5524,
      "step": 37
    },
    {
      "epoch": 0.48,
      "learning_rate": 9.967928531647374e-05,
      "loss": 0.5648,
      "step": 38
    },
    {
      "epoch": 0.49,
      "learning_rate": 9.966124079486872e-05,
      "loss": 0.4951,
      "step": 39
    },
    {
      "epoch": 0.51,
      "learning_rate": 9.96427041704248e-05,
      "loss": 0.5516,
      "step": 40
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.962367562682496e-05,
      "loss": 0.5582,
      "step": 41
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.960415535262671e-05,
      "loss": 0.5352,
      "step": 42
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.958414354126022e-05,
      "loss": 0.737,
      "step": 43
    },
    {
      "epoch": 0.56,
      "learning_rate": 9.956364039102642e-05,
      "loss": 0.4762,
      "step": 44
    },
    {
      "epoch": 0.57,
      "learning_rate": 9.954264610509497e-05,
      "loss": 0.4896,
      "step": 45
    },
    {
      "epoch": 0.58,
      "learning_rate": 9.952116089150232e-05,
      "loss": 0.4975,
      "step": 46
    },
    {
      "epoch": 0.59,
      "learning_rate": 9.94991849631496e-05,
      "loss": 0.5646,
      "step": 47
    },
    {
      "epoch": 0.61,
      "learning_rate": 9.947671853780054e-05,
      "loss": 0.5599,
      "step": 48
    },
    {
      "epoch": 0.62,
      "learning_rate": 9.94537618380793e-05,
      "loss": 0.5255,
      "step": 49
    },
    {
      "epoch": 0.63,
      "learning_rate": 9.943031509146825e-05,
      "loss": 0.5971,
      "step": 50
    },
    {
      "epoch": 0.65,
      "learning_rate": 9.940637853030572e-05,
      "loss": 0.5932,
      "step": 51
    },
    {
      "epoch": 0.66,
      "learning_rate": 9.938195239178374e-05,
      "loss": 0.5979,
      "step": 52
    },
    {
      "epoch": 0.67,
      "learning_rate": 9.935703691794565e-05,
      "loss": 0.496,
      "step": 53
    },
    {
      "epoch": 0.68,
      "learning_rate": 9.933163235568367e-05,
      "loss": 0.5341,
      "step": 54
    },
    {
      "epoch": 0.7,
      "learning_rate": 9.930573895673657e-05,
      "loss": 0.4661,
      "step": 55
    },
    {
      "epoch": 0.71,
      "learning_rate": 9.927935697768698e-05,
      "loss": 0.6222,
      "step": 56
    },
    {
      "epoch": 0.72,
      "learning_rate": 9.925248667995907e-05,
      "loss": 0.5015,
      "step": 57
    },
    {
      "epoch": 0.73,
      "learning_rate": 9.922512832981584e-05,
      "loss": 0.654,
      "step": 58
    },
    {
      "epoch": 0.75,
      "learning_rate": 9.919728219835643e-05,
      "loss": 0.4863,
      "step": 59
    },
    {
      "epoch": 0.76,
      "learning_rate": 9.916894856151357e-05,
      "loss": 0.6024,
      "step": 60
    },
    {
      "epoch": 0.77,
      "learning_rate": 9.914012770005072e-05,
      "loss": 0.4767,
      "step": 61
    },
    {
      "epoch": 0.78,
      "learning_rate": 9.91108198995594e-05,
      "loss": 0.5065,
      "step": 62
    },
    {
      "epoch": 0.8,
      "learning_rate": 9.908102545045625e-05,
      "loss": 0.6043,
      "step": 63
    },
    {
      "epoch": 0.81,
      "learning_rate": 9.905074464798024e-05,
      "loss": 0.6733,
      "step": 64
    },
    {
      "epoch": 0.82,
      "learning_rate": 9.901997779218967e-05,
      "loss": 0.522,
      "step": 65
    },
    {
      "epoch": 0.84,
      "learning_rate": 9.898872518795932e-05,
      "loss": 0.5205,
      "step": 66
    },
    {
      "epoch": 0.85,
      "learning_rate": 9.895698714497724e-05,
      "loss": 0.5216,
      "step": 67
    },
    {
      "epoch": 0.86,
      "learning_rate": 9.892476397774186e-05,
      "loss": 0.6014,
      "step": 68
    },
    {
      "epoch": 0.87,
      "learning_rate": 9.889205600555877e-05,
      "loss": 0.5321,
      "step": 69
    },
    {
      "epoch": 0.89,
      "learning_rate": 9.885886355253758e-05,
      "loss": 0.545,
      "step": 70
    },
    {
      "epoch": 0.9,
      "learning_rate": 9.882518694758875e-05,
      "loss": 0.4868,
      "step": 71
    },
    {
      "epoch": 0.91,
      "learning_rate": 9.879102652442024e-05,
      "loss": 0.5366,
      "step": 72
    },
    {
      "epoch": 0.92,
      "learning_rate": 9.875638262153431e-05,
      "loss": 0.5805,
      "step": 73
    },
    {
      "epoch": 0.94,
      "learning_rate": 9.872125558222409e-05,
      "loss": 0.585,
      "step": 74
    },
    {
      "epoch": 0.95,
      "learning_rate": 9.868564575457023e-05,
      "loss": 0.6058,
      "step": 75
    },
    {
      "epoch": 0.96,
      "learning_rate": 9.864955349143734e-05,
      "loss": 0.4402,
      "step": 76
    },
    {
      "epoch": 0.97,
      "learning_rate": 9.861297915047069e-05,
      "loss": 0.5349,
      "step": 77
    },
    {
      "epoch": 0.99,
      "learning_rate": 9.857592309409247e-05,
      "loss": 0.5443,
      "step": 78
    },
    {
      "epoch": 1.0,
      "learning_rate": 9.853838568949831e-05,
      "loss": 0.6198,
      "step": 79
    },
    {
      "epoch": 1.01,
      "learning_rate": 9.850036730865364e-05,
      "loss": 0.6008,
      "step": 80
    },
    {
      "epoch": 1.03,
      "learning_rate": 9.846186832828989e-05,
      "loss": 0.4168,
      "step": 81
    },
    {
      "epoch": 1.04,
      "learning_rate": 9.842288912990096e-05,
      "loss": 0.4738,
      "step": 82
    },
    {
      "epoch": 1.05,
      "learning_rate": 9.838343009973925e-05,
      "loss": 0.5146,
      "step": 83
    },
    {
      "epoch": 1.06,
      "learning_rate": 9.83434916288119e-05,
      "loss": 0.481,
      "step": 84
    },
    {
      "epoch": 1.08,
      "learning_rate": 9.830307411287695e-05,
      "loss": 0.5005,
      "step": 85
    },
    {
      "epoch": 1.09,
      "learning_rate": 9.82621779524394e-05,
      "loss": 0.4643,
      "step": 86
    },
    {
      "epoch": 1.1,
      "learning_rate": 9.822080355274719e-05,
      "loss": 0.5201,
      "step": 87
    },
    {
      "epoch": 1.11,
      "learning_rate": 9.817895132378725e-05,
      "loss": 0.4168,
      "step": 88
    },
    {
      "epoch": 1.13,
      "learning_rate": 9.813662168028144e-05,
      "loss": 0.4409,
      "step": 89
    },
    {
      "epoch": 1.14,
      "learning_rate": 9.809381504168234e-05,
      "loss": 0.4593,
      "step": 90
    },
    {
      "epoch": 1.15,
      "learning_rate": 9.805053183216923e-05,
      "loss": 0.4755,
      "step": 91
    },
    {
      "epoch": 1.16,
      "learning_rate": 9.800677248064382e-05,
      "loss": 0.4565,
      "step": 92
    },
    {
      "epoch": 1.18,
      "learning_rate": 9.796253742072596e-05,
      "loss": 0.4718,
      "step": 93
    },
    {
      "epoch": 1.19,
      "learning_rate": 9.791782709074944e-05,
      "loss": 0.4884,
      "step": 94
    },
    {
      "epoch": 1.2,
      "learning_rate": 9.787264193375753e-05,
      "loss": 0.5204,
      "step": 95
    },
    {
      "epoch": 1.22,
      "learning_rate": 9.782698239749873e-05,
      "loss": 0.4807,
      "step": 96
    },
    {
      "epoch": 1.23,
      "learning_rate": 9.778084893442218e-05,
      "loss": 0.4089,
      "step": 97
    },
    {
      "epoch": 1.24,
      "learning_rate": 9.77342420016733e-05,
      "loss": 0.4371,
      "step": 98
    },
    {
      "epoch": 1.25,
      "learning_rate": 9.768716206108921e-05,
      "loss": 0.4959,
      "step": 99
    },
    {
      "epoch": 1.27,
      "learning_rate": 9.763960957919413e-05,
      "loss": 0.4446,
      "step": 100
    },
    {
      "epoch": 1.28,
      "learning_rate": 9.759158502719481e-05,
      "loss": 0.4576,
      "step": 101
    },
    {
      "epoch": 1.29,
      "learning_rate": 9.754308888097583e-05,
      "loss": 0.3678,
      "step": 102
    },
    {
      "epoch": 1.3,
      "learning_rate": 9.749412162109485e-05,
      "loss": 0.3743,
      "step": 103
    },
    {
      "epoch": 1.32,
      "learning_rate": 9.744468373277797e-05,
      "loss": 0.4571,
      "step": 104
    },
    {
      "epoch": 1.33,
      "learning_rate": 9.739477570591473e-05,
      "loss": 0.3944,
      "step": 105
    },
    {
      "epoch": 1.34,
      "learning_rate": 9.734439803505345e-05,
      "loss": 0.3904,
      "step": 106
    },
    {
      "epoch": 1.35,
      "learning_rate": 9.729355121939621e-05,
      "loss": 0.494,
      "step": 107
    },
    {
      "epoch": 1.37,
      "learning_rate": 9.724223576279395e-05,
      "loss": 0.4904,
      "step": 108
    },
    {
      "epoch": 1.38,
      "learning_rate": 9.719045217374143e-05,
      "loss": 0.3727,
      "step": 109
    },
    {
      "epoch": 1.39,
      "learning_rate": 9.713820096537225e-05,
      "loss": 0.4916,
      "step": 110
    },
    {
      "epoch": 1.41,
      "learning_rate": 9.708548265545375e-05,
      "loss": 0.5386,
      "step": 111
    },
    {
      "epoch": 1.42,
      "learning_rate": 9.703229776638185e-05,
      "loss": 0.3815,
      "step": 112
    },
    {
      "epoch": 1.43,
      "learning_rate": 9.697864682517592e-05,
      "loss": 0.4243,
      "step": 113
    },
    {
      "epoch": 1.44,
      "learning_rate": 9.692453036347351e-05,
      "loss": 0.5205,
      "step": 114
    },
    {
      "epoch": 1.46,
      "learning_rate": 9.686994891752508e-05,
      "loss": 0.5158,
      "step": 115
    },
    {
      "epoch": 1.47,
      "learning_rate": 9.681490302818874e-05,
      "loss": 0.4424,
      "step": 116
    },
    {
      "epoch": 1.48,
      "learning_rate": 9.675939324092486e-05,
      "loss": 0.5825,
      "step": 117
    },
    {
      "epoch": 1.49,
      "learning_rate": 9.670342010579065e-05,
      "loss": 0.6082,
      "step": 118
    },
    {
      "epoch": 1.51,
      "learning_rate": 9.664698417743475e-05,
      "loss": 0.3913,
      "step": 119
    },
    {
      "epoch": 1.52,
      "learning_rate": 9.659008601509168e-05,
      "loss": 0.4178,
      "step": 120
    },
    {
      "epoch": 1.53,
      "learning_rate": 9.653272618257631e-05,
      "loss": 0.5543,
      "step": 121
    },
    {
      "epoch": 1.54,
      "learning_rate": 9.647490524827834e-05,
      "loss": 0.4127,
      "step": 122
    },
    {
      "epoch": 1.56,
      "learning_rate": 9.641662378515659e-05,
      "loss": 0.4079,
      "step": 123
    },
    {
      "epoch": 1.57,
      "learning_rate": 9.635788237073334e-05,
      "loss": 0.4811,
      "step": 124
    },
    {
      "epoch": 1.58,
      "learning_rate": 9.629868158708861e-05,
      "loss": 0.4229,
      "step": 125
    },
    {
      "epoch": 1.59,
      "learning_rate": 9.623902202085444e-05,
      "loss": 0.4332,
      "step": 126
    },
    {
      "epoch": 1.61,
      "learning_rate": 9.617890426320899e-05,
      "loss": 0.4451,
      "step": 127
    },
    {
      "epoch": 1.62,
      "learning_rate": 9.611832890987076e-05,
      "loss": 0.4018,
      "step": 128
    },
    {
      "epoch": 1.63,
      "learning_rate": 9.605729656109265e-05,
      "loss": 0.4175,
      "step": 129
    },
    {
      "epoch": 1.65,
      "learning_rate": 9.599580782165598e-05,
      "loss": 0.4841,
      "step": 130
    },
    {
      "epoch": 1.66,
      "learning_rate": 9.593386330086458e-05,
      "loss": 0.4353,
      "step": 131
    },
    {
      "epoch": 1.67,
      "learning_rate": 9.587146361253868e-05,
      "loss": 0.4262,
      "step": 132
    },
    {
      "epoch": 1.68,
      "learning_rate": 9.580860937500884e-05,
      "loss": 0.4572,
      "step": 133
    },
    {
      "epoch": 1.7,
      "learning_rate": 9.57453012111099e-05,
      "loss": 0.4856,
      "step": 134
    },
    {
      "epoch": 1.71,
      "learning_rate": 9.568153974817464e-05,
      "loss": 0.4451,
      "step": 135
    },
    {
      "epoch": 1.72,
      "learning_rate": 9.561732561802778e-05,
      "loss": 0.5691,
      "step": 136
    },
    {
      "epoch": 1.73,
      "learning_rate": 9.555265945697953e-05,
      "loss": 0.3989,
      "step": 137
    },
    {
      "epoch": 1.75,
      "learning_rate": 9.548754190581939e-05,
      "loss": 0.4935,
      "step": 138
    },
    {
      "epoch": 1.76,
      "learning_rate": 9.542197360980978e-05,
      "loss": 0.4637,
      "step": 139
    },
    {
      "epoch": 1.77,
      "learning_rate": 9.53559552186796e-05,
      "loss": 0.4895,
      "step": 140
    },
    {
      "epoch": 1.78,
      "learning_rate": 9.528948738661784e-05,
      "loss": 0.4625,
      "step": 141
    },
    {
      "epoch": 1.8,
      "learning_rate": 9.522257077226717e-05,
      "loss": 0.4409,
      "step": 142
    },
    {
      "epoch": 1.81,
      "learning_rate": 9.51552060387172e-05,
      "loss": 0.4123,
      "step": 143
    },
    {
      "epoch": 1.82,
      "learning_rate": 9.508739385349812e-05,
      "loss": 0.4801,
      "step": 144
    },
    {
      "epoch": 1.84,
      "learning_rate": 9.501913488857399e-05,
      "loss": 0.3965,
      "step": 145
    },
    {
      "epoch": 1.85,
      "learning_rate": 9.49504298203361e-05,
      "loss": 0.4027,
      "step": 146
    },
    {
      "epoch": 1.86,
      "learning_rate": 9.488127932959625e-05,
      "loss": 0.4779,
      "step": 147
    },
    {
      "epoch": 1.87,
      "learning_rate": 9.481168410158003e-05,
      "loss": 0.4168,
      "step": 148
    },
    {
      "epoch": 1.89,
      "learning_rate": 9.474164482592002e-05,
      "loss": 0.4352,
      "step": 149
    },
    {
      "epoch": 1.9,
      "learning_rate": 9.467116219664894e-05,
      "loss": 0.5395,
      "step": 150
    },
    {
      "epoch": 1.91,
      "learning_rate": 9.460023691219277e-05,
      "loss": 0.5874,
      "step": 151
    },
    {
      "epoch": 1.92,
      "learning_rate": 9.45288696753639e-05,
      "loss": 0.5014,
      "step": 152
    },
    {
      "epoch": 1.94,
      "learning_rate": 9.445706119335407e-05,
      "loss": 0.4106,
      "step": 153
    },
    {
      "epoch": 1.95,
      "learning_rate": 9.438481217772744e-05,
      "loss": 0.4717,
      "step": 154
    },
    {
      "epoch": 1.96,
      "learning_rate": 9.431212334441343e-05,
      "loss": 0.4105,
      "step": 155
    },
    {
      "epoch": 1.97,
      "learning_rate": 9.423899541369978e-05,
      "loss": 0.4421,
      "step": 156
    },
    {
      "epoch": 1.99,
      "learning_rate": 9.41654291102253e-05,
      "loss": 0.491,
      "step": 157
    },
    {
      "epoch": 2.0,
      "learning_rate": 9.409142516297269e-05,
      "loss": 0.4624,
      "step": 158
    },
    {
      "epoch": 2.01,
      "learning_rate": 9.401698430526142e-05,
      "loss": 0.2957,
      "step": 159
    },
    {
      "epoch": 2.03,
      "learning_rate": 9.394210727474028e-05,
      "loss": 0.3436,
      "step": 160
    },
    {
      "epoch": 2.04,
      "learning_rate": 9.386679481338033e-05,
      "loss": 0.3312,
      "step": 161
    },
    {
      "epoch": 2.05,
      "learning_rate": 9.379104766746722e-05,
      "loss": 0.3232,
      "step": 162
    },
    {
      "epoch": 2.06,
      "learning_rate": 9.371486658759416e-05,
      "loss": 0.3306,
      "step": 163
    },
    {
      "epoch": 2.08,
      "learning_rate": 9.363825232865413e-05,
      "loss": 0.3507,
      "step": 164
    },
    {
      "epoch": 2.09,
      "learning_rate": 9.356120564983266e-05,
      "loss": 0.3474,
      "step": 165
    },
    {
      "epoch": 2.1,
      "learning_rate": 9.348372731460023e-05,
      "loss": 0.3206,
      "step": 166
    },
    {
      "epoch": 2.11,
      "learning_rate": 9.340581809070459e-05,
      "loss": 0.2675,
      "step": 167
    },
    {
      "epoch": 2.13,
      "learning_rate": 9.332747875016332e-05,
      "loss": 0.3432,
      "step": 168
    },
    {
      "epoch": 2.14,
      "learning_rate": 9.324871006925613e-05,
      "loss": 0.3829,
      "step": 169
    },
    {
      "epoch": 2.15,
      "learning_rate": 9.316951282851707e-05,
      "loss": 0.3136,
      "step": 170
    },
    {
      "epoch": 2.16,
      "learning_rate": 9.308988781272694e-05,
      "loss": 0.3381,
      "step": 171
    },
    {
      "epoch": 2.18,
      "learning_rate": 9.300983581090541e-05,
      "loss": 0.3394,
      "step": 172
    },
    {
      "epoch": 2.19,
      "learning_rate": 9.292935761630326e-05,
      "loss": 0.3277,
      "step": 173
    },
    {
      "epoch": 2.2,
      "learning_rate": 9.284845402639446e-05,
      "loss": 0.3286,
      "step": 174
    },
    {
      "epoch": 2.22,
      "learning_rate": 9.276712584286833e-05,
      "loss": 0.3151,
      "step": 175
    },
    {
      "epoch": 2.23,
      "learning_rate": 9.26853738716216e-05,
      "loss": 0.3557,
      "step": 176
    },
    {
      "epoch": 2.24,
      "learning_rate": 9.260319892275034e-05,
      "loss": 0.2099,
      "step": 177
    },
    {
      "epoch": 2.25,
      "learning_rate": 9.2520601810542e-05,
      "loss": 0.3627,
      "step": 178
    },
    {
      "epoch": 2.27,
      "learning_rate": 9.243758335346735e-05,
      "loss": 0.285,
      "step": 179
    },
    {
      "epoch": 2.28,
      "learning_rate": 9.235414437417234e-05,
      "loss": 0.3189,
      "step": 180
    },
    {
      "epoch": 2.29,
      "learning_rate": 9.227028569946996e-05,
      "loss": 0.288,
      "step": 181
    },
    {
      "epoch": 2.3,
      "learning_rate": 9.2186008160332e-05,
      "loss": 0.2462,
      "step": 182
    },
    {
      "epoch": 2.32,
      "learning_rate": 9.210131259188095e-05,
      "loss": 0.2527,
      "step": 183
    },
    {
      "epoch": 2.33,
      "learning_rate": 9.201619983338153e-05,
      "loss": 0.3712,
      "step": 184
    },
    {
      "epoch": 2.34,
      "learning_rate": 9.193067072823251e-05,
      "loss": 0.3526,
      "step": 185
    },
    {
      "epoch": 2.35,
      "learning_rate": 9.18447261239584e-05,
      "loss": 0.2721,
      "step": 186
    },
    {
      "epoch": 2.37,
      "learning_rate": 9.175836687220084e-05,
      "loss": 0.2738,
      "step": 187
    },
    {
      "epoch": 2.38,
      "learning_rate": 9.167159382871039e-05,
      "loss": 0.3511,
      "step": 188
    },
    {
      "epoch": 2.39,
      "learning_rate": 9.15844078533379e-05,
      "loss": 0.3712,
      "step": 189
    },
    {
      "epoch": 2.41,
      "learning_rate": 9.149680981002609e-05,
      "loss": 0.3244,
      "step": 190
    },
    {
      "epoch": 2.42,
      "learning_rate": 9.140880056680088e-05,
      "loss": 0.3304,
      "step": 191
    },
    {
      "epoch": 2.43,
      "learning_rate": 9.13203809957629e-05,
      "loss": 0.3476,
      "step": 192
    },
    {
      "epoch": 2.44,
      "learning_rate": 9.123155197307876e-05,
      "loss": 0.2981,
      "step": 193
    },
    {
      "epoch": 2.46,
      "learning_rate": 9.114231437897244e-05,
      "loss": 0.2947,
      "step": 194
    },
    {
      "epoch": 2.47,
      "learning_rate": 9.105266909771653e-05,
      "loss": 0.2627,
      "step": 195
    },
    {
      "epoch": 2.48,
      "learning_rate": 9.096261701762342e-05,
      "loss": 0.306,
      "step": 196
    },
    {
      "epoch": 2.49,
      "learning_rate": 9.087215903103662e-05,
      "loss": 0.3326,
      "step": 197
    },
    {
      "epoch": 2.51,
      "learning_rate": 9.078129603432181e-05,
      "loss": 0.2348,
      "step": 198
    },
    {
      "epoch": 2.52,
      "learning_rate": 9.069002892785797e-05,
      "loss": 0.4223,
      "step": 199
    },
    {
      "epoch": 2.53,
      "learning_rate": 9.059835861602853e-05,
      "loss": 0.294,
      "step": 200
    },
    {
      "epoch": 2.54,
      "learning_rate": 9.050628600721234e-05,
      "loss": 0.269,
      "step": 201
    },
    {
      "epoch": 2.56,
      "learning_rate": 9.041381201377468e-05,
      "loss": 0.3375,
      "step": 202
    },
    {
      "epoch": 2.57,
      "learning_rate": 9.032093755205822e-05,
      "loss": 0.3022,
      "step": 203
    },
    {
      "epoch": 2.58,
      "learning_rate": 9.0227663542374e-05,
      "loss": 0.3297,
      "step": 204
    },
    {
      "epoch": 2.59,
      "learning_rate": 9.013399090899217e-05,
      "loss": 0.3108,
      "step": 205
    },
    {
      "epoch": 2.61,
      "learning_rate": 9.003992058013302e-05,
      "loss": 0.2948,
      "step": 206
    },
    {
      "epoch": 2.62,
      "learning_rate": 8.994545348795759e-05,
      "loss": 0.2719,
      "step": 207
    },
    {
      "epoch": 2.63,
      "learning_rate": 8.985059056855858e-05,
      "loss": 0.3116,
      "step": 208
    },
    {
      "epoch": 2.65,
      "learning_rate": 8.975533276195102e-05,
      "loss": 0.2567,
      "step": 209
    },
    {
      "epoch": 2.66,
      "learning_rate": 8.965968101206291e-05,
      "loss": 0.3751,
      "step": 210
    },
    {
      "epoch": 2.67,
      "learning_rate": 8.956363626672595e-05,
      "loss": 0.2538,
      "step": 211
    },
    {
      "epoch": 2.68,
      "learning_rate": 8.94671994776661e-05,
      "loss": 0.2973,
      "step": 212
    },
    {
      "epoch": 2.7,
      "learning_rate": 8.937037160049416e-05,
      "loss": 0.4477,
      "step": 213
    },
    {
      "epoch": 2.71,
      "learning_rate": 8.927315359469626e-05,
      "loss": 0.3517,
      "step": 214
    },
    {
      "epoch": 2.72,
      "learning_rate": 8.917554642362443e-05,
      "loss": 0.2779,
      "step": 215
    },
    {
      "epoch": 2.73,
      "learning_rate": 8.907755105448704e-05,
      "loss": 0.3196,
      "step": 216
    },
    {
      "epoch": 2.75,
      "learning_rate": 8.89791684583391e-05,
      "loss": 0.31,
      "step": 217
    },
    {
      "epoch": 2.76,
      "learning_rate": 8.888039961007282e-05,
      "loss": 0.2522,
      "step": 218
    },
    {
      "epoch": 2.77,
      "learning_rate": 8.87812454884078e-05,
      "loss": 0.2974,
      "step": 219
    },
    {
      "epoch": 2.78,
      "learning_rate": 8.868170707588142e-05,
      "loss": 0.2817,
      "step": 220
    },
    {
      "epoch": 2.8,
      "learning_rate": 8.858178535883905e-05,
      "loss": 0.3853,
      "step": 221
    },
    {
      "epoch": 2.81,
      "learning_rate": 8.848148132742431e-05,
      "loss": 0.2393,
      "step": 222
    },
    {
      "epoch": 2.82,
      "learning_rate": 8.838079597556925e-05,
      "loss": 0.3607,
      "step": 223
    },
    {
      "epoch": 2.84,
      "learning_rate": 8.827973030098448e-05,
      "loss": 0.3263,
      "step": 224
    },
    {
      "epoch": 2.85,
      "learning_rate": 8.81782853051493e-05,
      "loss": 0.3084,
      "step": 225
    },
    {
      "epoch": 2.86,
      "learning_rate": 8.807646199330187e-05,
      "loss": 0.3019,
      "step": 226
    },
    {
      "epoch": 2.87,
      "learning_rate": 8.797426137442897e-05,
      "loss": 0.3646,
      "step": 227
    },
    {
      "epoch": 2.89,
      "learning_rate": 8.787168446125638e-05,
      "loss": 0.3149,
      "step": 228
    },
    {
      "epoch": 2.9,
      "learning_rate": 8.776873227023852e-05,
      "loss": 0.326,
      "step": 229
    },
    {
      "epoch": 2.91,
      "learning_rate": 8.766540582154859e-05,
      "loss": 0.3216,
      "step": 230
    },
    {
      "epoch": 2.92,
      "learning_rate": 8.756170613906833e-05,
      "loss": 0.3348,
      "step": 231
    },
    {
      "epoch": 2.94,
      "learning_rate": 8.745763425037797e-05,
      "loss": 0.2524,
      "step": 232
    },
    {
      "epoch": 2.95,
      "learning_rate": 8.735319118674596e-05,
      "loss": 0.2725,
      "step": 233
    },
    {
      "epoch": 2.96,
      "learning_rate": 8.724837798311882e-05,
      "loss": 0.2733,
      "step": 234
    },
    {
      "epoch": 2.97,
      "learning_rate": 8.714319567811088e-05,
      "loss": 0.287,
      "step": 235
    },
    {
      "epoch": 2.99,
      "learning_rate": 8.703764531399392e-05,
      "loss": 0.2891,
      "step": 236
    },
    {
      "epoch": 3.0,
      "learning_rate": 8.69317279366869e-05,
      "loss": 0.2329,
      "step": 237
    },
    {
      "epoch": 3.01,
      "learning_rate": 8.682544459574562e-05,
      "loss": 0.1752,
      "step": 238
    },
    {
      "epoch": 3.03,
      "learning_rate": 8.671879634435224e-05,
      "loss": 0.1686,
      "step": 239
    },
    {
      "epoch": 3.04,
      "learning_rate": 8.661178423930491e-05,
      "loss": 0.2164,
      "step": 240
    },
    {
      "epoch": 3.05,
      "learning_rate": 8.650440934100728e-05,
      "loss": 0.1909,
      "step": 241
    },
    {
      "epoch": 3.06,
      "learning_rate": 8.639667271345798e-05,
      "loss": 0.1765,
      "step": 242
    },
    {
      "epoch": 3.08,
      "learning_rate": 8.628857542424009e-05,
      "loss": 0.1558,
      "step": 243
    },
    {
      "epoch": 3.09,
      "learning_rate": 8.618011854451056e-05,
      "loss": 0.1471,
      "step": 244
    },
    {
      "epoch": 3.1,
      "learning_rate": 8.607130314898956e-05,
      "loss": 0.175,
      "step": 245
    },
    {
      "epoch": 3.11,
      "learning_rate": 8.596213031594991e-05,
      "loss": 0.1692,
      "step": 246
    },
    {
      "epoch": 3.13,
      "learning_rate": 8.585260112720631e-05,
      "loss": 0.1176,
      "step": 247
    },
    {
      "epoch": 3.14,
      "learning_rate": 8.57427166681047e-05,
      "loss": 0.1773,
      "step": 248
    },
    {
      "epoch": 3.15,
      "learning_rate": 8.56324780275114e-05,
      "loss": 0.1821,
      "step": 249
    },
    {
      "epoch": 3.16,
      "learning_rate": 8.552188629780244e-05,
      "loss": 0.1471,
      "step": 250
    },
    {
      "epoch": 3.18,
      "learning_rate": 8.541094257485265e-05,
      "loss": 0.1949,
      "step": 251
    },
    {
      "epoch": 3.19,
      "learning_rate": 8.529964795802485e-05,
      "loss": 0.1234,
      "step": 252
    },
    {
      "epoch": 3.2,
      "learning_rate": 8.518800355015892e-05,
      "loss": 0.1625,
      "step": 253
    },
    {
      "epoch": 3.22,
      "learning_rate": 8.507601045756085e-05,
      "loss": 0.1417,
      "step": 254
    },
    {
      "epoch": 3.23,
      "learning_rate": 8.49636697899919e-05,
      "loss": 0.1088,
      "step": 255
    },
    {
      "epoch": 3.24,
      "learning_rate": 8.485098266065744e-05,
      "loss": 0.1524,
      "step": 256
    },
    {
      "epoch": 3.25,
      "learning_rate": 8.473795018619604e-05,
      "loss": 0.1711,
      "step": 257
    },
    {
      "epoch": 3.27,
      "learning_rate": 8.462457348666835e-05,
      "loss": 0.161,
      "step": 258
    },
    {
      "epoch": 3.28,
      "learning_rate": 8.4510853685546e-05,
      "loss": 0.1641,
      "step": 259
    },
    {
      "epoch": 3.29,
      "learning_rate": 8.439679190970052e-05,
      "loss": 0.1633,
      "step": 260
    },
    {
      "epoch": 3.3,
      "learning_rate": 8.428238928939207e-05,
      "loss": 0.1673,
      "step": 261
    },
    {
      "epoch": 3.32,
      "learning_rate": 8.416764695825835e-05,
      "loss": 0.1629,
      "step": 262
    },
    {
      "epoch": 3.33,
      "learning_rate": 8.405256605330331e-05,
      "loss": 0.177,
      "step": 263
    },
    {
      "epoch": 3.34,
      "learning_rate": 8.39371477148859e-05,
      "loss": 0.1188,
      "step": 264
    },
    {
      "epoch": 3.35,
      "learning_rate": 8.382139308670875e-05,
      "loss": 0.1934,
      "step": 265
    },
    {
      "epoch": 3.37,
      "learning_rate": 8.370530331580686e-05,
      "loss": 0.1049,
      "step": 266
    },
    {
      "epoch": 3.38,
      "learning_rate": 8.35888795525362e-05,
      "loss": 0.1376,
      "step": 267
    },
    {
      "epoch": 3.39,
      "learning_rate": 8.347212295056239e-05,
      "loss": 0.1574,
      "step": 268
    },
    {
      "epoch": 3.41,
      "learning_rate": 8.335503466684915e-05,
      "loss": 0.1222,
      "step": 269
    },
    {
      "epoch": 3.42,
      "learning_rate": 8.323761586164695e-05,
      "loss": 0.1634,
      "step": 270
    },
    {
      "epoch": 3.43,
      "learning_rate": 8.311986769848141e-05,
      "loss": 0.1764,
      "step": 271
    },
    {
      "epoch": 3.44,
      "learning_rate": 8.300179134414188e-05,
      "loss": 0.1621,
      "step": 272
    },
    {
      "epoch": 3.46,
      "learning_rate": 8.288338796866976e-05,
      "loss": 0.1549,
      "step": 273
    },
    {
      "epoch": 3.47,
      "learning_rate": 8.276465874534702e-05,
      "loss": 0.1053,
      "step": 274
    },
    {
      "epoch": 3.48,
      "learning_rate": 8.264560485068446e-05,
      "loss": 0.1224,
      "step": 275
    },
    {
      "epoch": 3.49,
      "learning_rate": 8.252622746441021e-05,
      "loss": 0.1361,
      "step": 276
    },
    {
      "epoch": 3.51,
      "learning_rate": 8.240652776945781e-05,
      "loss": 0.1753,
      "step": 277
    },
    {
      "epoch": 3.52,
      "learning_rate": 8.228650695195472e-05,
      "loss": 0.1539,
      "step": 278
    },
    {
      "epoch": 3.53,
      "learning_rate": 8.216616620121043e-05,
      "loss": 0.1603,
      "step": 279
    },
    {
      "epoch": 3.54,
      "learning_rate": 8.204550670970469e-05,
      "loss": 0.1519,
      "step": 280
    },
    {
      "epoch": 3.56,
      "learning_rate": 8.192452967307576e-05,
      "loss": 0.1494,
      "step": 281
    },
    {
      "epoch": 3.57,
      "learning_rate": 8.180323629010848e-05,
      "loss": 0.1695,
      "step": 282
    },
    {
      "epoch": 3.58,
      "learning_rate": 8.168162776272244e-05,
      "loss": 0.1384,
      "step": 283
    },
    {
      "epoch": 3.59,
      "learning_rate": 8.155970529596006e-05,
      "loss": 0.1552,
      "step": 284
    },
    {
      "epoch": 3.61,
      "learning_rate": 8.143747009797464e-05,
      "loss": 0.166,
      "step": 285
    },
    {
      "epoch": 3.62,
      "learning_rate": 8.131492338001839e-05,
      "loss": 0.2007,
      "step": 286
    },
    {
      "epoch": 3.63,
      "learning_rate": 8.119206635643045e-05,
      "loss": 0.1331,
      "step": 287
    },
    {
      "epoch": 3.65,
      "learning_rate": 8.106890024462481e-05,
      "loss": 0.2133,
      "step": 288
    },
    {
      "epoch": 3.66,
      "learning_rate": 8.094542626507828e-05,
      "loss": 0.1507,
      "step": 289
    },
    {
      "epoch": 3.67,
      "learning_rate": 8.082164564131845e-05,
      "loss": 0.1465,
      "step": 290
    },
    {
      "epoch": 3.68,
      "learning_rate": 8.069755959991142e-05,
      "loss": 0.1344,
      "step": 291
    },
    {
      "epoch": 3.7,
      "learning_rate": 8.057316937044977e-05,
      "loss": 0.1623,
      "step": 292
    },
    {
      "epoch": 3.71,
      "learning_rate": 8.044847618554034e-05,
      "loss": 0.1823,
      "step": 293
    },
    {
      "epoch": 3.72,
      "learning_rate": 8.032348128079203e-05,
      "loss": 0.1191,
      "step": 294
    },
    {
      "epoch": 3.73,
      "learning_rate": 8.019818589480352e-05,
      "loss": 0.1536,
      "step": 295
    },
    {
      "epoch": 3.75,
      "learning_rate": 8.0072591269151e-05,
      "loss": 0.1745,
      "step": 296
    },
    {
      "epoch": 3.76,
      "learning_rate": 7.994669864837594e-05,
      "loss": 0.1265,
      "step": 297
    },
    {
      "epoch": 3.77,
      "learning_rate": 7.982050927997264e-05,
      "loss": 0.1475,
      "step": 298
    },
    {
      "epoch": 3.78,
      "learning_rate": 7.969402441437594e-05,
      "loss": 0.1597,
      "step": 299
    },
    {
      "epoch": 3.8,
      "learning_rate": 7.956724530494887e-05,
      "loss": 0.1716,
      "step": 300
    },
    {
      "epoch": 3.81,
      "learning_rate": 7.944017320797013e-05,
      "loss": 0.1569,
      "step": 301
    },
    {
      "epoch": 3.82,
      "learning_rate": 7.931280938262169e-05,
      "loss": 0.1696,
      "step": 302
    },
    {
      "epoch": 3.84,
      "learning_rate": 7.918515509097634e-05,
      "loss": 0.1879,
      "step": 303
    },
    {
      "epoch": 3.85,
      "learning_rate": 7.905721159798513e-05,
      "loss": 0.1568,
      "step": 304
    },
    {
      "epoch": 3.86,
      "learning_rate": 7.89289801714649e-05,
      "loss": 0.1276,
      "step": 305
    },
    {
      "epoch": 3.87,
      "learning_rate": 7.880046208208563e-05,
      "loss": 0.1508,
      "step": 306
    },
    {
      "epoch": 3.89,
      "learning_rate": 7.867165860335792e-05,
      "loss": 0.1641,
      "step": 307
    },
    {
      "epoch": 3.9,
      "learning_rate": 7.854257101162037e-05,
      "loss": 0.1571,
      "step": 308
    },
    {
      "epoch": 3.91,
      "learning_rate": 7.841320058602688e-05,
      "loss": 0.1642,
      "step": 309
    },
    {
      "epoch": 3.92,
      "learning_rate": 7.828354860853399e-05,
      "loss": 0.2288,
      "step": 310
    },
    {
      "epoch": 3.94,
      "learning_rate": 7.815361636388827e-05,
      "loss": 0.1195,
      "step": 311
    },
    {
      "epoch": 3.95,
      "learning_rate": 7.802340513961342e-05,
      "loss": 0.1699,
      "step": 312
    },
    {
      "epoch": 3.96,
      "learning_rate": 7.789291622599767e-05,
      "loss": 0.1761,
      "step": 313
    },
    {
      "epoch": 3.97,
      "learning_rate": 7.776215091608085e-05,
      "loss": 0.2214,
      "step": 314
    },
    {
      "epoch": 3.99,
      "learning_rate": 7.763111050564178e-05,
      "loss": 0.2109,
      "step": 315
    },
    {
      "epoch": 4.0,
      "learning_rate": 7.749979629318516e-05,
      "loss": 0.1801,
      "step": 316
    },
    {
      "epoch": 4.01,
      "learning_rate": 7.736820957992895e-05,
      "loss": 0.0728,
      "step": 317
    },
    {
      "epoch": 4.03,
      "learning_rate": 7.723635166979133e-05,
      "loss": 0.0642,
      "step": 318
    },
    {
      "epoch": 4.04,
      "learning_rate": 7.710422386937784e-05,
      "loss": 0.0868,
      "step": 319
    },
    {
      "epoch": 4.05,
      "learning_rate": 7.697182748796841e-05,
      "loss": 0.0672,
      "step": 320
    },
    {
      "epoch": 4.06,
      "learning_rate": 7.683916383750436e-05,
      "loss": 0.0645,
      "step": 321
    },
    {
      "epoch": 4.08,
      "learning_rate": 7.670623423257548e-05,
      "loss": 0.0746,
      "step": 322
    },
    {
      "epoch": 4.09,
      "learning_rate": 7.657303999040693e-05,
      "loss": 0.0801,
      "step": 323
    },
    {
      "epoch": 4.1,
      "learning_rate": 7.64395824308462e-05,
      "loss": 0.0557,
      "step": 324
    },
    {
      "epoch": 4.11,
      "learning_rate": 7.630586287635008e-05,
      "loss": 0.061,
      "step": 325
    },
    {
      "epoch": 4.13,
      "learning_rate": 7.617188265197148e-05,
      "loss": 0.0666,
      "step": 326
    },
    {
      "epoch": 4.14,
      "learning_rate": 7.603764308534636e-05,
      "loss": 0.0495,
      "step": 327
    },
    {
      "epoch": 4.15,
      "learning_rate": 7.590314550668054e-05,
      "loss": 0.0591,
      "step": 328
    },
    {
      "epoch": 4.16,
      "learning_rate": 7.576839124873653e-05,
      "loss": 0.0914,
      "step": 329
    },
    {
      "epoch": 4.18,
      "learning_rate": 7.563338164682036e-05,
      "loss": 0.1023,
      "step": 330
    },
    {
      "epoch": 4.19,
      "learning_rate": 7.549811803876825e-05,
      "loss": 0.1043,
      "step": 331
    },
    {
      "epoch": 4.2,
      "learning_rate": 7.536260176493348e-05,
      "loss": 0.054,
      "step": 332
    },
    {
      "epoch": 4.22,
      "learning_rate": 7.5226834168173e-05,
      "loss": 0.0712,
      "step": 333
    },
    {
      "epoch": 4.23,
      "learning_rate": 7.509081659383417e-05,
      "loss": 0.0814,
      "step": 334
    },
    {
      "epoch": 4.24,
      "learning_rate": 7.495455038974146e-05,
      "loss": 0.0573,
      "step": 335
    },
    {
      "epoch": 4.25,
      "learning_rate": 7.481803690618304e-05,
      "loss": 0.0603,
      "step": 336
    },
    {
      "epoch": 4.27,
      "learning_rate": 7.46812774958974e-05,
      "loss": 0.0565,
      "step": 337
    },
    {
      "epoch": 4.28,
      "learning_rate": 7.454427351405999e-05,
      "loss": 0.063,
      "step": 338
    },
    {
      "epoch": 4.29,
      "learning_rate": 7.440702631826977e-05,
      "loss": 0.0666,
      "step": 339
    },
    {
      "epoch": 4.3,
      "learning_rate": 7.426953726853574e-05,
      "loss": 0.0669,
      "step": 340
    },
    {
      "epoch": 4.32,
      "learning_rate": 7.413180772726348e-05,
      "loss": 0.0812,
      "step": 341
    },
    {
      "epoch": 4.33,
      "learning_rate": 7.399383905924165e-05,
      "loss": 0.081,
      "step": 342
    },
    {
      "epoch": 4.34,
      "learning_rate": 7.385563263162847e-05,
      "loss": 0.0697,
      "step": 343
    },
    {
      "epoch": 4.35,
      "learning_rate": 7.371718981393815e-05,
      "loss": 0.0639,
      "step": 344
    },
    {
      "epoch": 4.37,
      "learning_rate": 7.357851197802735e-05,
      "loss": 0.0792,
      "step": 345
    },
    {
      "epoch": 4.38,
      "learning_rate": 7.343960049808156e-05,
      "loss": 0.0665,
      "step": 346
    },
    {
      "epoch": 4.39,
      "learning_rate": 7.330045675060149e-05,
      "loss": 0.0593,
      "step": 347
    },
    {
      "epoch": 4.41,
      "learning_rate": 7.316108211438945e-05,
      "loss": 0.0307,
      "step": 348
    },
    {
      "epoch": 4.42,
      "learning_rate": 7.302147797053569e-05,
      "loss": 0.0656,
      "step": 349
    },
    {
      "epoch": 4.43,
      "learning_rate": 7.288164570240463e-05,
      "loss": 0.049,
      "step": 350
    },
    {
      "epoch": 4.44,
      "learning_rate": 7.274158669562126e-05,
      "loss": 0.0653,
      "step": 351
    },
    {
      "epoch": 4.46,
      "learning_rate": 7.26013023380574e-05,
      "loss": 0.0668,
      "step": 352
    },
    {
      "epoch": 4.47,
      "learning_rate": 7.246079401981784e-05,
      "loss": 0.0894,
      "step": 353
    },
    {
      "epoch": 4.48,
      "learning_rate": 7.232006313322667e-05,
      "loss": 0.0641,
      "step": 354
    },
    {
      "epoch": 4.49,
      "learning_rate": 7.217911107281352e-05,
      "loss": 0.0905,
      "step": 355
    },
    {
      "epoch": 4.51,
      "learning_rate": 7.203793923529956e-05,
      "loss": 0.0753,
      "step": 356
    },
    {
      "epoch": 4.52,
      "learning_rate": 7.189654901958385e-05,
      "loss": 0.0737,
      "step": 357
    },
    {
      "epoch": 4.53,
      "learning_rate": 7.175494182672939e-05,
      "loss": 0.0828,
      "step": 358
    },
    {
      "epoch": 4.54,
      "learning_rate": 7.161311905994922e-05,
      "loss": 0.0597,
      "step": 359
    },
    {
      "epoch": 4.56,
      "learning_rate": 7.147108212459257e-05,
      "loss": 0.0623,
      "step": 360
    },
    {
      "epoch": 4.57,
      "learning_rate": 7.13288324281309e-05,
      "loss": 0.0751,
      "step": 361
    },
    {
      "epoch": 4.58,
      "learning_rate": 7.118637138014396e-05,
      "loss": 0.0586,
      "step": 362
    },
    {
      "epoch": 4.59,
      "learning_rate": 7.104370039230583e-05,
      "loss": 0.0794,
      "step": 363
    },
    {
      "epoch": 4.61,
      "learning_rate": 7.090082087837092e-05,
      "loss": 0.0554,
      "step": 364
    },
    {
      "epoch": 4.62,
      "learning_rate": 7.075773425415994e-05,
      "loss": 0.0601,
      "step": 365
    },
    {
      "epoch": 4.63,
      "learning_rate": 7.061444193754596e-05,
      "loss": 0.0867,
      "step": 366
    },
    {
      "epoch": 4.65,
      "learning_rate": 7.047094534844023e-05,
      "loss": 0.0828,
      "step": 367
    },
    {
      "epoch": 4.66,
      "learning_rate": 7.032724590877821e-05,
      "loss": 0.0879,
      "step": 368
    },
    {
      "epoch": 4.67,
      "learning_rate": 7.018334504250545e-05,
      "loss": 0.068,
      "step": 369
    },
    {
      "epoch": 4.68,
      "learning_rate": 7.003924417556343e-05,
      "loss": 0.0586,
      "step": 370
    },
    {
      "epoch": 4.7,
      "learning_rate": 6.989494473587554e-05,
      "loss": 0.0971,
      "step": 371
    },
    {
      "epoch": 4.71,
      "learning_rate": 6.975044815333282e-05,
      "loss": 0.0864,
      "step": 372
    },
    {
      "epoch": 4.72,
      "learning_rate": 6.960575585977984e-05,
      "loss": 0.0482,
      "step": 373
    },
    {
      "epoch": 4.73,
      "learning_rate": 6.946086928900054e-05,
      "loss": 0.0812,
      "step": 374
    },
    {
      "epoch": 4.75,
      "learning_rate": 6.931578987670396e-05,
      "loss": 0.0756,
      "step": 375
    },
    {
      "epoch": 4.76,
      "learning_rate": 6.917051906051006e-05,
      "loss": 0.0585,
      "step": 376
    },
    {
      "epoch": 4.77,
      "learning_rate": 6.902505827993541e-05,
      "loss": 0.0653,
      "step": 377
    },
    {
      "epoch": 4.78,
      "learning_rate": 6.887940897637908e-05,
      "loss": 0.0783,
      "step": 378
    },
    {
      "epoch": 4.8,
      "learning_rate": 6.873357259310815e-05,
      "loss": 0.0849,
      "step": 379
    },
    {
      "epoch": 4.81,
      "learning_rate": 6.858755057524354e-05,
      "loss": 0.0642,
      "step": 380
    },
    {
      "epoch": 4.82,
      "learning_rate": 6.844134436974567e-05,
      "loss": 0.0651,
      "step": 381
    },
    {
      "epoch": 4.84,
      "learning_rate": 6.829495542540013e-05,
      "loss": 0.0511,
      "step": 382
    },
    {
      "epoch": 4.85,
      "learning_rate": 6.814838519280324e-05,
      "loss": 0.0713,
      "step": 383
    },
    {
      "epoch": 4.86,
      "learning_rate": 6.80016351243478e-05,
      "loss": 0.0853,
      "step": 384
    },
    {
      "epoch": 4.87,
      "learning_rate": 6.785470667420862e-05,
      "loss": 0.0777,
      "step": 385
    },
    {
      "epoch": 4.89,
      "learning_rate": 6.77076012983281e-05,
      "loss": 0.0714,
      "step": 386
    },
    {
      "epoch": 4.9,
      "learning_rate": 6.75603204544019e-05,
      "loss": 0.051,
      "step": 387
    },
    {
      "epoch": 4.91,
      "learning_rate": 6.741286560186437e-05,
      "loss": 0.0831,
      "step": 388
    },
    {
      "epoch": 4.92,
      "learning_rate": 6.726523820187413e-05,
      "loss": 0.0613,
      "step": 389
    },
    {
      "epoch": 4.94,
      "learning_rate": 6.711743971729967e-05,
      "loss": 0.092,
      "step": 390
    },
    {
      "epoch": 4.95,
      "learning_rate": 6.696947161270476e-05,
      "loss": 0.0457,
      "step": 391
    },
    {
      "epoch": 4.96,
      "learning_rate": 6.682133535433393e-05,
      "loss": 0.0491,
      "step": 392
    },
    {
      "epoch": 4.97,
      "learning_rate": 6.667303241009803e-05,
      "loss": 0.0788,
      "step": 393
    },
    {
      "epoch": 4.99,
      "learning_rate": 6.652456424955963e-05,
      "loss": 0.0598,
      "step": 394
    },
    {
      "epoch": 5.0,
      "learning_rate": 6.637593234391843e-05,
      "loss": 0.0446,
      "step": 395
    },
    {
      "epoch": 5.01,
      "learning_rate": 6.622713816599673e-05,
      "loss": 0.02,
      "step": 396
    },
    {
      "epoch": 5.03,
      "learning_rate": 6.60781831902248e-05,
      "loss": 0.03,
      "step": 397
    },
    {
      "epoch": 5.04,
      "learning_rate": 6.592906889262632e-05,
      "loss": 0.0296,
      "step": 398
    },
    {
      "epoch": 5.05,
      "learning_rate": 6.577979675080369e-05,
      "loss": 0.0267,
      "step": 399
    },
    {
      "epoch": 5.06,
      "learning_rate": 6.563036824392344e-05,
      "loss": 0.0358,
      "step": 400
    },
    {
      "epoch": 5.08,
      "learning_rate": 6.548078485270152e-05,
      "loss": 0.0387,
      "step": 401
    },
    {
      "epoch": 5.09,
      "learning_rate": 6.533104805938873e-05,
      "loss": 0.0335,
      "step": 402
    },
    {
      "epoch": 5.1,
      "learning_rate": 6.518115934775585e-05,
      "loss": 0.0275,
      "step": 403
    },
    {
      "epoch": 5.11,
      "learning_rate": 6.503112020307916e-05,
      "loss": 0.0258,
      "step": 404
    },
    {
      "epoch": 5.13,
      "learning_rate": 6.488093211212555e-05,
      "loss": 0.0534,
      "step": 405
    },
    {
      "epoch": 5.14,
      "learning_rate": 6.473059656313782e-05,
      "loss": 0.0176,
      "step": 406
    },
    {
      "epoch": 5.15,
      "learning_rate": 6.458011504582005e-05,
      "loss": 0.025,
      "step": 407
    },
    {
      "epoch": 5.16,
      "learning_rate": 6.442948905132266e-05,
      "loss": 0.0172,
      "step": 408
    },
    {
      "epoch": 5.18,
      "learning_rate": 6.427872007222777e-05,
      "loss": 0.0312,
      "step": 409
    },
    {
      "epoch": 5.19,
      "learning_rate": 6.412780960253436e-05,
      "loss": 0.0279,
      "step": 410
    },
    {
      "epoch": 5.2,
      "learning_rate": 6.397675913764347e-05,
      "loss": 0.0279,
      "step": 411
    },
    {
      "epoch": 5.22,
      "learning_rate": 6.382557017434332e-05,
      "loss": 0.0198,
      "step": 412
    },
    {
      "epoch": 5.23,
      "learning_rate": 6.367424421079463e-05,
      "loss": 0.0404,
      "step": 413
    },
    {
      "epoch": 5.24,
      "learning_rate": 6.352278274651561e-05,
      "loss": 0.0266,
      "step": 414
    },
    {
      "epoch": 5.25,
      "learning_rate": 6.337118728236721e-05,
      "loss": 0.0341,
      "step": 415
    },
    {
      "epoch": 5.27,
      "learning_rate": 6.321945932053822e-05,
      "loss": 0.0205,
      "step": 416
    },
    {
      "epoch": 5.28,
      "learning_rate": 6.306760036453035e-05,
      "loss": 0.0341,
      "step": 417
    },
    {
      "epoch": 5.29,
      "learning_rate": 6.291561191914333e-05,
      "loss": 0.0398,
      "step": 418
    },
    {
      "epoch": 5.3,
      "learning_rate": 6.276349549046007e-05,
      "loss": 0.0287,
      "step": 419
    },
    {
      "epoch": 5.32,
      "learning_rate": 6.261125258583171e-05,
      "loss": 0.0491,
      "step": 420
    },
    {
      "epoch": 5.33,
      "learning_rate": 6.245888471386263e-05,
      "loss": 0.0235,
      "step": 421
    },
    {
      "epoch": 5.34,
      "learning_rate": 6.230639338439549e-05,
      "loss": 0.0312,
      "step": 422
    },
    {
      "epoch": 5.35,
      "learning_rate": 6.215378010849641e-05,
      "loss": 0.0413,
      "step": 423
    },
    {
      "epoch": 5.37,
      "learning_rate": 6.200104639843985e-05,
      "loss": 0.0316,
      "step": 424
    },
    {
      "epoch": 5.38,
      "learning_rate": 6.184819376769364e-05,
      "loss": 0.044,
      "step": 425
    },
    {
      "epoch": 5.39,
      "learning_rate": 6.169522373090412e-05,
      "loss": 0.0233,
      "step": 426
    },
    {
      "epoch": 5.41,
      "learning_rate": 6.154213780388092e-05,
      "loss": 0.0152,
      "step": 427
    },
    {
      "epoch": 5.42,
      "learning_rate": 6.138893750358212e-05,
      "loss": 0.0589,
      "step": 428
    },
    {
      "epoch": 5.43,
      "learning_rate": 6.123562434809912e-05,
      "loss": 0.0534,
      "step": 429
    },
    {
      "epoch": 5.44,
      "learning_rate": 6.108219985664161e-05,
      "loss": 0.0453,
      "step": 430
    },
    {
      "epoch": 5.46,
      "learning_rate": 6.0928665549522554e-05,
      "loss": 0.0252,
      "step": 431
    },
    {
      "epoch": 5.47,
      "learning_rate": 6.0775022948143115e-05,
      "loss": 0.0337,
      "step": 432
    },
    {
      "epoch": 5.48,
      "learning_rate": 6.06212735749775e-05,
      "loss": 0.0406,
      "step": 433
    },
    {
      "epoch": 5.49,
      "learning_rate": 6.046741895355802e-05,
      "loss": 0.044,
      "step": 434
    },
    {
      "epoch": 5.51,
      "learning_rate": 6.031346060845986e-05,
      "loss": 0.0646,
      "step": 435
    },
    {
      "epoch": 5.52,
      "learning_rate": 6.015940006528602e-05,
      "loss": 0.0238,
      "step": 436
    },
    {
      "epoch": 5.53,
      "learning_rate": 6.0005238850652234e-05,
      "loss": 0.0517,
      "step": 437
    },
    {
      "epoch": 5.54,
      "learning_rate": 5.9850978492171794e-05,
      "loss": 0.0274,
      "step": 438
    },
    {
      "epoch": 5.56,
      "learning_rate": 5.96966205184404e-05,
      "loss": 0.0247,
      "step": 439
    },
    {
      "epoch": 5.57,
      "learning_rate": 5.954216645902109e-05,
      "loss": 0.0648,
      "step": 440
    },
    {
      "epoch": 5.58,
      "learning_rate": 5.9387617844429e-05,
      "loss": 0.0282,
      "step": 441
    },
    {
      "epoch": 5.59,
      "learning_rate": 5.923297620611623e-05,
      "loss": 0.0537,
      "step": 442
    },
    {
      "epoch": 5.61,
      "learning_rate": 5.907824307645669e-05,
      "loss": 0.048,
      "step": 443
    },
    {
      "epoch": 5.62,
      "learning_rate": 5.892341998873089e-05,
      "loss": 0.0227,
      "step": 444
    },
    {
      "epoch": 5.63,
      "learning_rate": 5.876850847711073e-05,
      "loss": 0.0211,
      "step": 445
    },
    {
      "epoch": 5.65,
      "learning_rate": 5.861351007664434e-05,
      "loss": 0.0318,
      "step": 446
    },
    {
      "epoch": 5.66,
      "learning_rate": 5.845842632324088e-05,
      "loss": 0.0225,
      "step": 447
    },
    {
      "epoch": 5.67,
      "learning_rate": 5.83032587536552e-05,
      "loss": 0.0482,
      "step": 448
    },
    {
      "epoch": 5.68,
      "learning_rate": 5.814800890547278e-05,
      "loss": 0.0543,
      "step": 449
    },
    {
      "epoch": 5.7,
      "learning_rate": 5.799267831709442e-05,
      "loss": 0.0264,
      "step": 450
    },
    {
      "epoch": 5.71,
      "learning_rate": 5.78372685277209e-05,
      "loss": 0.0311,
      "step": 451
    },
    {
      "epoch": 5.72,
      "learning_rate": 5.7681781077337905e-05,
      "loss": 0.0524,
      "step": 452
    },
    {
      "epoch": 5.73,
      "learning_rate": 5.752621750670068e-05,
      "loss": 0.0186,
      "step": 453
    },
    {
      "epoch": 5.75,
      "learning_rate": 5.737057935731868e-05,
      "loss": 0.0269,
      "step": 454
    },
    {
      "epoch": 5.76,
      "learning_rate": 5.721486817144044e-05,
      "loss": 0.0312,
      "step": 455
    },
    {
      "epoch": 5.77,
      "learning_rate": 5.705908549203823e-05,
      "loss": 0.0356,
      "step": 456
    },
    {
      "epoch": 5.78,
      "learning_rate": 5.690323286279274e-05,
      "loss": 0.0507,
      "step": 457
    },
    {
      "epoch": 5.8,
      "learning_rate": 5.674731182807781e-05,
      "loss": 0.0399,
      "step": 458
    },
    {
      "epoch": 5.81,
      "learning_rate": 5.659132393294514e-05,
      "loss": 0.0364,
      "step": 459
    },
    {
      "epoch": 5.82,
      "learning_rate": 5.643527072310891e-05,
      "loss": 0.0453,
      "step": 460
    },
    {
      "epoch": 5.84,
      "learning_rate": 5.627915374493061e-05,
      "loss": 0.021,
      "step": 461
    },
    {
      "epoch": 5.85,
      "learning_rate": 5.612297454540352e-05,
      "loss": 0.0155,
      "step": 462
    },
    {
      "epoch": 5.86,
      "learning_rate": 5.596673467213756e-05,
      "loss": 0.0336,
      "step": 463
    },
    {
      "epoch": 5.87,
      "learning_rate": 5.581043567334383e-05,
      "loss": 0.0245,
      "step": 464
    },
    {
      "epoch": 5.89,
      "learning_rate": 5.5654079097819345e-05,
      "loss": 0.0359,
      "step": 465
    },
    {
      "epoch": 5.9,
      "learning_rate": 5.5497666494931654e-05,
      "loss": 0.0245,
      "step": 466
    },
    {
      "epoch": 5.91,
      "learning_rate": 5.5341199414603493e-05,
      "loss": 0.0246,
      "step": 467
    },
    {
      "epoch": 5.92,
      "learning_rate": 5.518467940729739e-05,
      "loss": 0.0222,
      "step": 468
    },
    {
      "epoch": 5.94,
      "learning_rate": 5.502810802400039e-05,
      "loss": 0.0317,
      "step": 469
    },
    {
      "epoch": 5.95,
      "learning_rate": 5.487148681620862e-05,
      "loss": 0.0199,
      "step": 470
    },
    {
      "epoch": 5.96,
      "learning_rate": 5.4714817335911894e-05,
      "loss": 0.0559,
      "step": 471
    },
    {
      "epoch": 5.97,
      "learning_rate": 5.455810113557839e-05,
      "loss": 0.0405,
      "step": 472
    },
    {
      "epoch": 5.99,
      "learning_rate": 5.440133976813926e-05,
      "loss": 0.0416,
      "step": 473
    },
    {
      "epoch": 6.0,
      "learning_rate": 5.4244534786973214e-05,
      "loss": 0.0208,
      "step": 474
    },
    {
      "epoch": 6.01,
      "learning_rate": 5.40876877458911e-05,
      "loss": 0.0295,
      "step": 475
    },
    {
      "epoch": 6.03,
      "learning_rate": 5.3930800199120616e-05,
      "loss": 0.0106,
      "step": 476
    },
    {
      "epoch": 6.04,
      "learning_rate": 5.377387370129079e-05,
      "loss": 0.0165,
      "step": 477
    },
    {
      "epoch": 6.05,
      "learning_rate": 5.361690980741663e-05,
      "loss": 0.0098,
      "step": 478
    },
    {
      "epoch": 6.06,
      "learning_rate": 5.345991007288371e-05,
      "loss": 0.0259,
      "step": 479
    },
    {
      "epoch": 6.08,
      "learning_rate": 5.330287605343279e-05,
      "loss": 0.0065,
      "step": 480
    },
    {
      "epoch": 6.09,
      "learning_rate": 5.314580930514431e-05,
      "loss": 0.0162,
      "step": 481
    },
    {
      "epoch": 6.1,
      "learning_rate": 5.298871138442307e-05,
      "loss": 0.0107,
      "step": 482
    },
    {
      "epoch": 6.11,
      "learning_rate": 5.283158384798275e-05,
      "loss": 0.0126,
      "step": 483
    },
    {
      "epoch": 6.13,
      "learning_rate": 5.267442825283048e-05,
      "loss": 0.0275,
      "step": 484
    },
    {
      "epoch": 6.14,
      "learning_rate": 5.2517246156251455e-05,
      "loss": 0.022,
      "step": 485
    },
    {
      "epoch": 6.15,
      "learning_rate": 5.236003911579345e-05,
      "loss": 0.0101,
      "step": 486
    },
    {
      "epoch": 6.16,
      "learning_rate": 5.220280868925145e-05,
      "loss": 0.0189,
      "step": 487
    },
    {
      "epoch": 6.18,
      "learning_rate": 5.204555643465215e-05,
      "loss": 0.0205,
      "step": 488
    },
    {
      "epoch": 6.19,
      "learning_rate": 5.1888283910238555e-05,
      "loss": 0.0123,
      "step": 489
    },
    {
      "epoch": 6.2,
      "learning_rate": 5.173099267445451e-05,
      "loss": 0.015,
      "step": 490
    },
    {
      "epoch": 6.22,
      "learning_rate": 5.157368428592933e-05,
      "loss": 0.0116,
      "step": 491
    },
    {
      "epoch": 6.23,
      "learning_rate": 5.1416360303462206e-05,
      "loss": 0.0219,
      "step": 492
    },
    {
      "epoch": 6.24,
      "learning_rate": 5.125902228600693e-05,
      "loss": 0.0095,
      "step": 493
    },
    {
      "epoch": 6.25,
      "learning_rate": 5.110167179265636e-05,
      "loss": 0.0287,
      "step": 494
    },
    {
      "epoch": 6.27,
      "learning_rate": 5.094431038262693e-05,
      "loss": 0.0184,
      "step": 495
    },
    {
      "epoch": 6.28,
      "learning_rate": 5.078693961524329e-05,
      "loss": 0.0118,
      "step": 496
    },
    {
      "epoch": 6.29,
      "learning_rate": 5.062956104992285e-05,
      "loss": 0.0147,
      "step": 497
    },
    {
      "epoch": 6.3,
      "learning_rate": 5.0472176246160184e-05,
      "loss": 0.02,
      "step": 498
    },
    {
      "epoch": 6.32,
      "learning_rate": 5.031478676351179e-05,
      "loss": 0.0126,
      "step": 499
    },
    {
      "epoch": 6.33,
      "learning_rate": 5.01573941615805e-05,
      "loss": 0.0118,
      "step": 500
    }
  ],
  "max_steps": 1000,
  "num_train_epochs": 13,
  "total_flos": 3.637158979269427e+16,
  "trial_name": null,
  "trial_params": null
}