{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.225806451612903,
  "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": 1.0268,
      "step": 1
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0001,
      "loss": 1.1517,
      "step": 2
    },
    {
      "epoch": 0.02,
      "learning_rate": 9.999975227016531e-05,
      "loss": 0.9465,
      "step": 3
    },
    {
      "epoch": 0.03,
      "learning_rate": 9.999900908311602e-05,
      "loss": 1.0734,
      "step": 4
    },
    {
      "epoch": 0.03,
      "learning_rate": 9.999777044621652e-05,
      "loss": 0.9898,
      "step": 5
    },
    {
      "epoch": 0.04,
      "learning_rate": 9.999603637174071e-05,
      "loss": 1.1077,
      "step": 6
    },
    {
      "epoch": 0.05,
      "learning_rate": 9.999380687687188e-05,
      "loss": 1.156,
      "step": 7
    },
    {
      "epoch": 0.05,
      "learning_rate": 9.999108198370249e-05,
      "loss": 1.0887,
      "step": 8
    },
    {
      "epoch": 0.06,
      "learning_rate": 9.998786171923407e-05,
      "loss": 0.8221,
      "step": 9
    },
    {
      "epoch": 0.06,
      "learning_rate": 9.998414611537681e-05,
      "loss": 0.8151,
      "step": 10
    },
    {
      "epoch": 0.07,
      "learning_rate": 9.997993520894937e-05,
      "loss": 0.8667,
      "step": 11
    },
    {
      "epoch": 0.08,
      "learning_rate": 9.997522904167844e-05,
      "loss": 0.8528,
      "step": 12
    },
    {
      "epoch": 0.08,
      "learning_rate": 9.997002766019832e-05,
      "loss": 0.9431,
      "step": 13
    },
    {
      "epoch": 0.09,
      "learning_rate": 9.996433111605052e-05,
      "loss": 0.9324,
      "step": 14
    },
    {
      "epoch": 0.1,
      "learning_rate": 9.99581394656832e-05,
      "loss": 0.8742,
      "step": 15
    },
    {
      "epoch": 0.1,
      "learning_rate": 9.995145277045061e-05,
      "loss": 1.0315,
      "step": 16
    },
    {
      "epoch": 0.11,
      "learning_rate": 9.994427109661253e-05,
      "loss": 0.6145,
      "step": 17
    },
    {
      "epoch": 0.12,
      "learning_rate": 9.993659451533353e-05,
      "loss": 0.6864,
      "step": 18
    },
    {
      "epoch": 0.12,
      "learning_rate": 9.992842310268233e-05,
      "loss": 0.8386,
      "step": 19
    },
    {
      "epoch": 0.13,
      "learning_rate": 9.991975693963107e-05,
      "loss": 0.6645,
      "step": 20
    },
    {
      "epoch": 0.14,
      "learning_rate": 9.99105961120544e-05,
      "loss": 0.7871,
      "step": 21
    },
    {
      "epoch": 0.14,
      "learning_rate": 9.990094071072877e-05,
      "loss": 0.913,
      "step": 22
    },
    {
      "epoch": 0.15,
      "learning_rate": 9.989079083133139e-05,
      "loss": 0.7164,
      "step": 23
    },
    {
      "epoch": 0.15,
      "learning_rate": 9.988014657443941e-05,
      "loss": 0.7411,
      "step": 24
    },
    {
      "epoch": 0.16,
      "learning_rate": 9.986900804552878e-05,
      "loss": 0.7257,
      "step": 25
    },
    {
      "epoch": 0.17,
      "learning_rate": 9.985737535497337e-05,
      "loss": 0.7692,
      "step": 26
    },
    {
      "epoch": 0.17,
      "learning_rate": 9.984524861804376e-05,
      "loss": 0.63,
      "step": 27
    },
    {
      "epoch": 0.18,
      "learning_rate": 9.983262795490613e-05,
      "loss": 0.7015,
      "step": 28
    },
    {
      "epoch": 0.19,
      "learning_rate": 9.981951349062106e-05,
      "loss": 0.6968,
      "step": 29
    },
    {
      "epoch": 0.19,
      "learning_rate": 9.980590535514233e-05,
      "loss": 0.7845,
      "step": 30
    },
    {
      "epoch": 0.2,
      "learning_rate": 9.979180368331558e-05,
      "loss": 0.7577,
      "step": 31
    },
    {
      "epoch": 0.21,
      "learning_rate": 9.9777208614877e-05,
      "loss": 0.7563,
      "step": 32
    },
    {
      "epoch": 0.21,
      "learning_rate": 9.976212029445194e-05,
      "loss": 0.6315,
      "step": 33
    },
    {
      "epoch": 0.22,
      "learning_rate": 9.97465388715535e-05,
      "loss": 0.7325,
      "step": 34
    },
    {
      "epoch": 0.23,
      "learning_rate": 9.9730464500581e-05,
      "loss": 0.6608,
      "step": 35
    },
    {
      "epoch": 0.23,
      "learning_rate": 9.971389734081848e-05,
      "loss": 0.6093,
      "step": 36
    },
    {
      "epoch": 0.24,
      "learning_rate": 9.969683755643317e-05,
      "loss": 0.6489,
      "step": 37
    },
    {
      "epoch": 0.25,
      "learning_rate": 9.967928531647374e-05,
      "loss": 0.6433,
      "step": 38
    },
    {
      "epoch": 0.25,
      "learning_rate": 9.966124079486872e-05,
      "loss": 0.4883,
      "step": 39
    },
    {
      "epoch": 0.26,
      "learning_rate": 9.96427041704248e-05,
      "loss": 0.5188,
      "step": 40
    },
    {
      "epoch": 0.26,
      "learning_rate": 9.962367562682496e-05,
      "loss": 0.711,
      "step": 41
    },
    {
      "epoch": 0.27,
      "learning_rate": 9.960415535262671e-05,
      "loss": 0.5365,
      "step": 42
    },
    {
      "epoch": 0.28,
      "learning_rate": 9.958414354126022e-05,
      "loss": 0.7529,
      "step": 43
    },
    {
      "epoch": 0.28,
      "learning_rate": 9.956364039102642e-05,
      "loss": 0.464,
      "step": 44
    },
    {
      "epoch": 0.29,
      "learning_rate": 9.954264610509497e-05,
      "loss": 0.6099,
      "step": 45
    },
    {
      "epoch": 0.3,
      "learning_rate": 9.952116089150232e-05,
      "loss": 0.7049,
      "step": 46
    },
    {
      "epoch": 0.3,
      "learning_rate": 9.94991849631496e-05,
      "loss": 0.6887,
      "step": 47
    },
    {
      "epoch": 0.31,
      "learning_rate": 9.947671853780054e-05,
      "loss": 0.6614,
      "step": 48
    },
    {
      "epoch": 0.32,
      "learning_rate": 9.94537618380793e-05,
      "loss": 0.5701,
      "step": 49
    },
    {
      "epoch": 0.32,
      "learning_rate": 9.943031509146825e-05,
      "loss": 0.7035,
      "step": 50
    },
    {
      "epoch": 0.33,
      "learning_rate": 9.940637853030572e-05,
      "loss": 0.7875,
      "step": 51
    },
    {
      "epoch": 0.34,
      "learning_rate": 9.938195239178374e-05,
      "loss": 0.5118,
      "step": 52
    },
    {
      "epoch": 0.34,
      "learning_rate": 9.935703691794565e-05,
      "loss": 0.6286,
      "step": 53
    },
    {
      "epoch": 0.35,
      "learning_rate": 9.933163235568367e-05,
      "loss": 0.6611,
      "step": 54
    },
    {
      "epoch": 0.35,
      "learning_rate": 9.930573895673657e-05,
      "loss": 0.5909,
      "step": 55
    },
    {
      "epoch": 0.36,
      "learning_rate": 9.927935697768698e-05,
      "loss": 0.6,
      "step": 56
    },
    {
      "epoch": 0.37,
      "learning_rate": 9.925248667995907e-05,
      "loss": 0.6585,
      "step": 57
    },
    {
      "epoch": 0.37,
      "learning_rate": 9.922512832981584e-05,
      "loss": 0.8174,
      "step": 58
    },
    {
      "epoch": 0.38,
      "learning_rate": 9.919728219835643e-05,
      "loss": 0.4368,
      "step": 59
    },
    {
      "epoch": 0.39,
      "learning_rate": 9.916894856151357e-05,
      "loss": 0.5216,
      "step": 60
    },
    {
      "epoch": 0.39,
      "learning_rate": 9.914012770005072e-05,
      "loss": 0.7608,
      "step": 61
    },
    {
      "epoch": 0.4,
      "learning_rate": 9.91108198995594e-05,
      "loss": 0.4761,
      "step": 62
    },
    {
      "epoch": 0.41,
      "learning_rate": 9.908102545045625e-05,
      "loss": 0.664,
      "step": 63
    },
    {
      "epoch": 0.41,
      "learning_rate": 9.905074464798024e-05,
      "loss": 0.5137,
      "step": 64
    },
    {
      "epoch": 0.42,
      "learning_rate": 9.901997779218967e-05,
      "loss": 0.6164,
      "step": 65
    },
    {
      "epoch": 0.43,
      "learning_rate": 9.898872518795932e-05,
      "loss": 0.5914,
      "step": 66
    },
    {
      "epoch": 0.43,
      "learning_rate": 9.895698714497724e-05,
      "loss": 0.6915,
      "step": 67
    },
    {
      "epoch": 0.44,
      "learning_rate": 9.892476397774186e-05,
      "loss": 0.4589,
      "step": 68
    },
    {
      "epoch": 0.45,
      "learning_rate": 9.889205600555877e-05,
      "loss": 0.6707,
      "step": 69
    },
    {
      "epoch": 0.45,
      "learning_rate": 9.885886355253758e-05,
      "loss": 0.5891,
      "step": 70
    },
    {
      "epoch": 0.46,
      "learning_rate": 9.882518694758875e-05,
      "loss": 0.6385,
      "step": 71
    },
    {
      "epoch": 0.46,
      "learning_rate": 9.879102652442024e-05,
      "loss": 0.7422,
      "step": 72
    },
    {
      "epoch": 0.47,
      "learning_rate": 9.875638262153431e-05,
      "loss": 0.6817,
      "step": 73
    },
    {
      "epoch": 0.48,
      "learning_rate": 9.872125558222409e-05,
      "loss": 0.5551,
      "step": 74
    },
    {
      "epoch": 0.48,
      "learning_rate": 9.868564575457023e-05,
      "loss": 0.6608,
      "step": 75
    },
    {
      "epoch": 0.49,
      "learning_rate": 9.864955349143734e-05,
      "loss": 0.6215,
      "step": 76
    },
    {
      "epoch": 0.5,
      "learning_rate": 9.861297915047069e-05,
      "loss": 0.6019,
      "step": 77
    },
    {
      "epoch": 0.5,
      "learning_rate": 9.857592309409247e-05,
      "loss": 0.5369,
      "step": 78
    },
    {
      "epoch": 0.51,
      "learning_rate": 9.853838568949831e-05,
      "loss": 0.7186,
      "step": 79
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.850036730865364e-05,
      "loss": 0.6195,
      "step": 80
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.846186832828989e-05,
      "loss": 0.7495,
      "step": 81
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.842288912990096e-05,
      "loss": 0.7342,
      "step": 82
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.838343009973925e-05,
      "loss": 0.543,
      "step": 83
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.83434916288119e-05,
      "loss": 0.5429,
      "step": 84
    },
    {
      "epoch": 0.55,
      "learning_rate": 9.830307411287695e-05,
      "loss": 0.7106,
      "step": 85
    },
    {
      "epoch": 0.55,
      "learning_rate": 9.82621779524394e-05,
      "loss": 0.6062,
      "step": 86
    },
    {
      "epoch": 0.56,
      "learning_rate": 9.822080355274719e-05,
      "loss": 0.5878,
      "step": 87
    },
    {
      "epoch": 0.57,
      "learning_rate": 9.817895132378725e-05,
      "loss": 0.5608,
      "step": 88
    },
    {
      "epoch": 0.57,
      "learning_rate": 9.813662168028144e-05,
      "loss": 0.499,
      "step": 89
    },
    {
      "epoch": 0.58,
      "learning_rate": 9.809381504168234e-05,
      "loss": 0.6296,
      "step": 90
    },
    {
      "epoch": 0.59,
      "learning_rate": 9.805053183216923e-05,
      "loss": 0.4594,
      "step": 91
    },
    {
      "epoch": 0.59,
      "learning_rate": 9.800677248064382e-05,
      "loss": 0.6637,
      "step": 92
    },
    {
      "epoch": 0.6,
      "learning_rate": 9.796253742072596e-05,
      "loss": 0.5033,
      "step": 93
    },
    {
      "epoch": 0.61,
      "learning_rate": 9.791782709074944e-05,
      "loss": 0.7305,
      "step": 94
    },
    {
      "epoch": 0.61,
      "learning_rate": 9.787264193375753e-05,
      "loss": 0.6021,
      "step": 95
    },
    {
      "epoch": 0.62,
      "learning_rate": 9.782698239749873e-05,
      "loss": 0.531,
      "step": 96
    },
    {
      "epoch": 0.63,
      "learning_rate": 9.778084893442218e-05,
      "loss": 0.5017,
      "step": 97
    },
    {
      "epoch": 0.63,
      "learning_rate": 9.77342420016733e-05,
      "loss": 0.6915,
      "step": 98
    },
    {
      "epoch": 0.64,
      "learning_rate": 9.768716206108921e-05,
      "loss": 0.4774,
      "step": 99
    },
    {
      "epoch": 0.65,
      "learning_rate": 9.763960957919413e-05,
      "loss": 0.502,
      "step": 100
    },
    {
      "epoch": 0.65,
      "learning_rate": 9.759158502719481e-05,
      "loss": 0.5251,
      "step": 101
    },
    {
      "epoch": 0.66,
      "learning_rate": 9.754308888097583e-05,
      "loss": 0.7634,
      "step": 102
    },
    {
      "epoch": 0.66,
      "learning_rate": 9.749412162109485e-05,
      "loss": 0.7938,
      "step": 103
    },
    {
      "epoch": 0.67,
      "learning_rate": 9.744468373277797e-05,
      "loss": 0.625,
      "step": 104
    },
    {
      "epoch": 0.68,
      "learning_rate": 9.739477570591473e-05,
      "loss": 0.701,
      "step": 105
    },
    {
      "epoch": 0.68,
      "learning_rate": 9.734439803505345e-05,
      "loss": 0.5768,
      "step": 106
    },
    {
      "epoch": 0.69,
      "learning_rate": 9.729355121939621e-05,
      "loss": 0.5662,
      "step": 107
    },
    {
      "epoch": 0.7,
      "learning_rate": 9.724223576279395e-05,
      "loss": 0.4875,
      "step": 108
    },
    {
      "epoch": 0.7,
      "learning_rate": 9.719045217374143e-05,
      "loss": 0.4934,
      "step": 109
    },
    {
      "epoch": 0.71,
      "learning_rate": 9.713820096537225e-05,
      "loss": 0.4517,
      "step": 110
    },
    {
      "epoch": 0.72,
      "learning_rate": 9.708548265545375e-05,
      "loss": 0.7039,
      "step": 111
    },
    {
      "epoch": 0.72,
      "learning_rate": 9.703229776638185e-05,
      "loss": 0.6789,
      "step": 112
    },
    {
      "epoch": 0.73,
      "learning_rate": 9.697864682517592e-05,
      "loss": 0.5693,
      "step": 113
    },
    {
      "epoch": 0.74,
      "learning_rate": 9.692453036347351e-05,
      "loss": 0.6704,
      "step": 114
    },
    {
      "epoch": 0.74,
      "learning_rate": 9.686994891752508e-05,
      "loss": 0.8254,
      "step": 115
    },
    {
      "epoch": 0.75,
      "learning_rate": 9.681490302818874e-05,
      "loss": 0.6009,
      "step": 116
    },
    {
      "epoch": 0.75,
      "learning_rate": 9.675939324092486e-05,
      "loss": 0.9807,
      "step": 117
    },
    {
      "epoch": 0.76,
      "learning_rate": 9.670342010579065e-05,
      "loss": 0.5363,
      "step": 118
    },
    {
      "epoch": 0.77,
      "learning_rate": 9.664698417743475e-05,
      "loss": 0.5341,
      "step": 119
    },
    {
      "epoch": 0.77,
      "learning_rate": 9.659008601509168e-05,
      "loss": 0.5928,
      "step": 120
    },
    {
      "epoch": 0.78,
      "learning_rate": 9.653272618257631e-05,
      "loss": 0.6052,
      "step": 121
    },
    {
      "epoch": 0.79,
      "learning_rate": 9.647490524827834e-05,
      "loss": 0.6151,
      "step": 122
    },
    {
      "epoch": 0.79,
      "learning_rate": 9.641662378515659e-05,
      "loss": 0.5901,
      "step": 123
    },
    {
      "epoch": 0.8,
      "learning_rate": 9.635788237073334e-05,
      "loss": 0.5193,
      "step": 124
    },
    {
      "epoch": 0.81,
      "learning_rate": 9.629868158708861e-05,
      "loss": 0.5785,
      "step": 125
    },
    {
      "epoch": 0.81,
      "learning_rate": 9.623902202085444e-05,
      "loss": 0.4423,
      "step": 126
    },
    {
      "epoch": 0.82,
      "learning_rate": 9.617890426320899e-05,
      "loss": 0.4556,
      "step": 127
    },
    {
      "epoch": 0.83,
      "learning_rate": 9.611832890987076e-05,
      "loss": 0.6943,
      "step": 128
    },
    {
      "epoch": 0.83,
      "learning_rate": 9.605729656109265e-05,
      "loss": 0.8507,
      "step": 129
    },
    {
      "epoch": 0.84,
      "learning_rate": 9.599580782165598e-05,
      "loss": 0.5016,
      "step": 130
    },
    {
      "epoch": 0.85,
      "learning_rate": 9.593386330086458e-05,
      "loss": 0.5495,
      "step": 131
    },
    {
      "epoch": 0.85,
      "learning_rate": 9.587146361253868e-05,
      "loss": 0.5518,
      "step": 132
    },
    {
      "epoch": 0.86,
      "learning_rate": 9.580860937500884e-05,
      "loss": 0.4475,
      "step": 133
    },
    {
      "epoch": 0.86,
      "learning_rate": 9.57453012111099e-05,
      "loss": 0.7087,
      "step": 134
    },
    {
      "epoch": 0.87,
      "learning_rate": 9.568153974817464e-05,
      "loss": 0.4885,
      "step": 135
    },
    {
      "epoch": 0.88,
      "learning_rate": 9.561732561802778e-05,
      "loss": 0.7467,
      "step": 136
    },
    {
      "epoch": 0.88,
      "learning_rate": 9.555265945697953e-05,
      "loss": 0.6204,
      "step": 137
    },
    {
      "epoch": 0.89,
      "learning_rate": 9.548754190581939e-05,
      "loss": 0.4334,
      "step": 138
    },
    {
      "epoch": 0.9,
      "learning_rate": 9.542197360980978e-05,
      "loss": 0.5795,
      "step": 139
    },
    {
      "epoch": 0.9,
      "learning_rate": 9.53559552186796e-05,
      "loss": 0.6091,
      "step": 140
    },
    {
      "epoch": 0.91,
      "learning_rate": 9.528948738661784e-05,
      "loss": 0.5542,
      "step": 141
    },
    {
      "epoch": 0.92,
      "learning_rate": 9.522257077226717e-05,
      "loss": 0.649,
      "step": 142
    },
    {
      "epoch": 0.92,
      "learning_rate": 9.51552060387172e-05,
      "loss": 0.6229,
      "step": 143
    },
    {
      "epoch": 0.93,
      "learning_rate": 9.508739385349812e-05,
      "loss": 0.7629,
      "step": 144
    },
    {
      "epoch": 0.94,
      "learning_rate": 9.501913488857399e-05,
      "loss": 0.736,
      "step": 145
    },
    {
      "epoch": 0.94,
      "learning_rate": 9.49504298203361e-05,
      "loss": 0.5591,
      "step": 146
    },
    {
      "epoch": 0.95,
      "learning_rate": 9.488127932959625e-05,
      "loss": 0.5068,
      "step": 147
    },
    {
      "epoch": 0.95,
      "learning_rate": 9.481168410158003e-05,
      "loss": 0.634,
      "step": 148
    },
    {
      "epoch": 0.96,
      "learning_rate": 9.474164482592002e-05,
      "loss": 0.5849,
      "step": 149
    },
    {
      "epoch": 0.97,
      "learning_rate": 9.467116219664894e-05,
      "loss": 0.6225,
      "step": 150
    },
    {
      "epoch": 0.97,
      "learning_rate": 9.460023691219277e-05,
      "loss": 0.6337,
      "step": 151
    },
    {
      "epoch": 0.98,
      "learning_rate": 9.45288696753639e-05,
      "loss": 0.5001,
      "step": 152
    },
    {
      "epoch": 0.99,
      "learning_rate": 9.445706119335407e-05,
      "loss": 0.5411,
      "step": 153
    },
    {
      "epoch": 0.99,
      "learning_rate": 9.438481217772744e-05,
      "loss": 0.5927,
      "step": 154
    },
    {
      "epoch": 1.0,
      "learning_rate": 9.431212334441343e-05,
      "loss": 0.5507,
      "step": 155
    },
    {
      "epoch": 1.01,
      "learning_rate": 9.423899541369978e-05,
      "loss": 0.6211,
      "step": 156
    },
    {
      "epoch": 1.01,
      "learning_rate": 9.41654291102253e-05,
      "loss": 0.5246,
      "step": 157
    },
    {
      "epoch": 1.02,
      "learning_rate": 9.409142516297269e-05,
      "loss": 0.5259,
      "step": 158
    },
    {
      "epoch": 1.03,
      "learning_rate": 9.401698430526142e-05,
      "loss": 0.5211,
      "step": 159
    },
    {
      "epoch": 1.03,
      "learning_rate": 9.394210727474028e-05,
      "loss": 0.4185,
      "step": 160
    },
    {
      "epoch": 1.04,
      "learning_rate": 9.386679481338033e-05,
      "loss": 0.4048,
      "step": 161
    },
    {
      "epoch": 1.05,
      "learning_rate": 9.379104766746722e-05,
      "loss": 0.4784,
      "step": 162
    },
    {
      "epoch": 1.05,
      "learning_rate": 9.371486658759416e-05,
      "loss": 0.668,
      "step": 163
    },
    {
      "epoch": 1.06,
      "learning_rate": 9.363825232865413e-05,
      "loss": 0.4245,
      "step": 164
    },
    {
      "epoch": 1.06,
      "learning_rate": 9.356120564983266e-05,
      "loss": 0.4072,
      "step": 165
    },
    {
      "epoch": 1.07,
      "learning_rate": 9.348372731460023e-05,
      "loss": 0.4508,
      "step": 166
    },
    {
      "epoch": 1.08,
      "learning_rate": 9.340581809070459e-05,
      "loss": 0.5202,
      "step": 167
    },
    {
      "epoch": 1.08,
      "learning_rate": 9.332747875016332e-05,
      "loss": 0.5079,
      "step": 168
    },
    {
      "epoch": 1.09,
      "learning_rate": 9.324871006925613e-05,
      "loss": 0.5253,
      "step": 169
    },
    {
      "epoch": 1.1,
      "learning_rate": 9.316951282851707e-05,
      "loss": 0.4746,
      "step": 170
    },
    {
      "epoch": 1.1,
      "learning_rate": 9.308988781272694e-05,
      "loss": 0.445,
      "step": 171
    },
    {
      "epoch": 1.11,
      "learning_rate": 9.300983581090541e-05,
      "loss": 0.4136,
      "step": 172
    },
    {
      "epoch": 1.12,
      "learning_rate": 9.292935761630326e-05,
      "loss": 0.6599,
      "step": 173
    },
    {
      "epoch": 1.12,
      "learning_rate": 9.284845402639446e-05,
      "loss": 0.5104,
      "step": 174
    },
    {
      "epoch": 1.13,
      "learning_rate": 9.276712584286833e-05,
      "loss": 0.4904,
      "step": 175
    },
    {
      "epoch": 1.14,
      "learning_rate": 9.26853738716216e-05,
      "loss": 0.447,
      "step": 176
    },
    {
      "epoch": 1.14,
      "learning_rate": 9.260319892275034e-05,
      "loss": 0.4087,
      "step": 177
    },
    {
      "epoch": 1.15,
      "learning_rate": 9.2520601810542e-05,
      "loss": 0.492,
      "step": 178
    },
    {
      "epoch": 1.15,
      "learning_rate": 9.243758335346735e-05,
      "loss": 0.4826,
      "step": 179
    },
    {
      "epoch": 1.16,
      "learning_rate": 9.235414437417234e-05,
      "loss": 0.4253,
      "step": 180
    },
    {
      "epoch": 1.17,
      "learning_rate": 9.227028569946996e-05,
      "loss": 0.4347,
      "step": 181
    },
    {
      "epoch": 1.17,
      "learning_rate": 9.2186008160332e-05,
      "loss": 0.5012,
      "step": 182
    },
    {
      "epoch": 1.18,
      "learning_rate": 9.210131259188095e-05,
      "loss": 0.4528,
      "step": 183
    },
    {
      "epoch": 1.19,
      "learning_rate": 9.201619983338153e-05,
      "loss": 0.5214,
      "step": 184
    },
    {
      "epoch": 1.19,
      "learning_rate": 9.193067072823251e-05,
      "loss": 0.5223,
      "step": 185
    },
    {
      "epoch": 1.2,
      "learning_rate": 9.18447261239584e-05,
      "loss": 0.4836,
      "step": 186
    },
    {
      "epoch": 1.21,
      "learning_rate": 9.175836687220084e-05,
      "loss": 0.4164,
      "step": 187
    },
    {
      "epoch": 1.21,
      "learning_rate": 9.167159382871039e-05,
      "loss": 0.5401,
      "step": 188
    },
    {
      "epoch": 1.22,
      "learning_rate": 9.15844078533379e-05,
      "loss": 0.5124,
      "step": 189
    },
    {
      "epoch": 1.23,
      "learning_rate": 9.149680981002609e-05,
      "loss": 0.4166,
      "step": 190
    },
    {
      "epoch": 1.23,
      "learning_rate": 9.140880056680088e-05,
      "loss": 0.347,
      "step": 191
    },
    {
      "epoch": 1.24,
      "learning_rate": 9.13203809957629e-05,
      "loss": 0.5943,
      "step": 192
    },
    {
      "epoch": 1.25,
      "learning_rate": 9.123155197307876e-05,
      "loss": 0.5109,
      "step": 193
    },
    {
      "epoch": 1.25,
      "learning_rate": 9.114231437897244e-05,
      "loss": 0.4884,
      "step": 194
    },
    {
      "epoch": 1.26,
      "learning_rate": 9.105266909771653e-05,
      "loss": 0.5741,
      "step": 195
    },
    {
      "epoch": 1.26,
      "learning_rate": 9.096261701762342e-05,
      "loss": 0.4132,
      "step": 196
    },
    {
      "epoch": 1.27,
      "learning_rate": 9.087215903103662e-05,
      "loss": 0.4921,
      "step": 197
    },
    {
      "epoch": 1.28,
      "learning_rate": 9.078129603432181e-05,
      "loss": 0.5276,
      "step": 198
    },
    {
      "epoch": 1.28,
      "learning_rate": 9.069002892785797e-05,
      "loss": 0.5683,
      "step": 199
    },
    {
      "epoch": 1.29,
      "learning_rate": 9.059835861602853e-05,
      "loss": 0.5686,
      "step": 200
    },
    {
      "epoch": 1.3,
      "learning_rate": 9.050628600721234e-05,
      "loss": 0.5042,
      "step": 201
    },
    {
      "epoch": 1.3,
      "learning_rate": 9.041381201377468e-05,
      "loss": 0.3473,
      "step": 202
    },
    {
      "epoch": 1.31,
      "learning_rate": 9.032093755205822e-05,
      "loss": 0.4839,
      "step": 203
    },
    {
      "epoch": 1.32,
      "learning_rate": 9.0227663542374e-05,
      "loss": 0.5131,
      "step": 204
    },
    {
      "epoch": 1.32,
      "learning_rate": 9.013399090899217e-05,
      "loss": 0.4379,
      "step": 205
    },
    {
      "epoch": 1.33,
      "learning_rate": 9.003992058013302e-05,
      "loss": 0.3901,
      "step": 206
    },
    {
      "epoch": 1.34,
      "learning_rate": 8.994545348795759e-05,
      "loss": 0.7088,
      "step": 207
    },
    {
      "epoch": 1.34,
      "learning_rate": 8.985059056855858e-05,
      "loss": 0.6866,
      "step": 208
    },
    {
      "epoch": 1.35,
      "learning_rate": 8.975533276195102e-05,
      "loss": 0.5805,
      "step": 209
    },
    {
      "epoch": 1.35,
      "learning_rate": 8.965968101206291e-05,
      "loss": 0.5227,
      "step": 210
    },
    {
      "epoch": 1.36,
      "learning_rate": 8.956363626672595e-05,
      "loss": 0.6249,
      "step": 211
    },
    {
      "epoch": 1.37,
      "learning_rate": 8.94671994776661e-05,
      "loss": 0.3752,
      "step": 212
    },
    {
      "epoch": 1.37,
      "learning_rate": 8.937037160049416e-05,
      "loss": 0.4021,
      "step": 213
    },
    {
      "epoch": 1.38,
      "learning_rate": 8.927315359469626e-05,
      "loss": 0.5918,
      "step": 214
    },
    {
      "epoch": 1.39,
      "learning_rate": 8.917554642362443e-05,
      "loss": 0.7731,
      "step": 215
    },
    {
      "epoch": 1.39,
      "learning_rate": 8.907755105448704e-05,
      "loss": 0.48,
      "step": 216
    },
    {
      "epoch": 1.4,
      "learning_rate": 8.89791684583391e-05,
      "loss": 0.6369,
      "step": 217
    },
    {
      "epoch": 1.41,
      "learning_rate": 8.888039961007282e-05,
      "loss": 0.3829,
      "step": 218
    },
    {
      "epoch": 1.41,
      "learning_rate": 8.87812454884078e-05,
      "loss": 0.64,
      "step": 219
    },
    {
      "epoch": 1.42,
      "learning_rate": 8.868170707588142e-05,
      "loss": 0.4687,
      "step": 220
    },
    {
      "epoch": 1.43,
      "learning_rate": 8.858178535883905e-05,
      "loss": 0.4399,
      "step": 221
    },
    {
      "epoch": 1.43,
      "learning_rate": 8.848148132742431e-05,
      "loss": 0.5487,
      "step": 222
    },
    {
      "epoch": 1.44,
      "learning_rate": 8.838079597556925e-05,
      "loss": 0.4131,
      "step": 223
    },
    {
      "epoch": 1.45,
      "learning_rate": 8.827973030098448e-05,
      "loss": 0.5555,
      "step": 224
    },
    {
      "epoch": 1.45,
      "learning_rate": 8.81782853051493e-05,
      "loss": 0.3712,
      "step": 225
    },
    {
      "epoch": 1.46,
      "learning_rate": 8.807646199330187e-05,
      "loss": 0.5312,
      "step": 226
    },
    {
      "epoch": 1.46,
      "learning_rate": 8.797426137442897e-05,
      "loss": 0.3888,
      "step": 227
    },
    {
      "epoch": 1.47,
      "learning_rate": 8.787168446125638e-05,
      "loss": 0.49,
      "step": 228
    },
    {
      "epoch": 1.48,
      "learning_rate": 8.776873227023852e-05,
      "loss": 0.5249,
      "step": 229
    },
    {
      "epoch": 1.48,
      "learning_rate": 8.766540582154859e-05,
      "loss": 0.4472,
      "step": 230
    },
    {
      "epoch": 1.49,
      "learning_rate": 8.756170613906833e-05,
      "loss": 0.4636,
      "step": 231
    },
    {
      "epoch": 1.5,
      "learning_rate": 8.745763425037797e-05,
      "loss": 0.5517,
      "step": 232
    },
    {
      "epoch": 1.5,
      "learning_rate": 8.735319118674596e-05,
      "loss": 0.4611,
      "step": 233
    },
    {
      "epoch": 1.51,
      "learning_rate": 8.724837798311882e-05,
      "loss": 0.5853,
      "step": 234
    },
    {
      "epoch": 1.52,
      "learning_rate": 8.714319567811088e-05,
      "loss": 0.3383,
      "step": 235
    },
    {
      "epoch": 1.52,
      "learning_rate": 8.703764531399392e-05,
      "loss": 0.6781,
      "step": 236
    },
    {
      "epoch": 1.53,
      "learning_rate": 8.69317279366869e-05,
      "loss": 0.4439,
      "step": 237
    },
    {
      "epoch": 1.54,
      "learning_rate": 8.682544459574562e-05,
      "loss": 0.488,
      "step": 238
    },
    {
      "epoch": 1.54,
      "learning_rate": 8.671879634435224e-05,
      "loss": 0.4344,
      "step": 239
    },
    {
      "epoch": 1.55,
      "learning_rate": 8.661178423930491e-05,
      "loss": 0.3145,
      "step": 240
    },
    {
      "epoch": 1.55,
      "learning_rate": 8.650440934100728e-05,
      "loss": 0.4356,
      "step": 241
    },
    {
      "epoch": 1.56,
      "learning_rate": 8.639667271345798e-05,
      "loss": 0.5421,
      "step": 242
    },
    {
      "epoch": 1.57,
      "learning_rate": 8.628857542424009e-05,
      "loss": 0.4701,
      "step": 243
    },
    {
      "epoch": 1.57,
      "learning_rate": 8.618011854451056e-05,
      "loss": 0.4127,
      "step": 244
    },
    {
      "epoch": 1.58,
      "learning_rate": 8.607130314898956e-05,
      "loss": 0.5318,
      "step": 245
    },
    {
      "epoch": 1.59,
      "learning_rate": 8.596213031594991e-05,
      "loss": 0.5162,
      "step": 246
    },
    {
      "epoch": 1.59,
      "learning_rate": 8.585260112720631e-05,
      "loss": 0.5307,
      "step": 247
    },
    {
      "epoch": 1.6,
      "learning_rate": 8.57427166681047e-05,
      "loss": 0.4985,
      "step": 248
    },
    {
      "epoch": 1.61,
      "learning_rate": 8.56324780275114e-05,
      "loss": 0.4366,
      "step": 249
    },
    {
      "epoch": 1.61,
      "learning_rate": 8.552188629780244e-05,
      "loss": 0.5154,
      "step": 250
    },
    {
      "epoch": 1.62,
      "learning_rate": 8.541094257485265e-05,
      "loss": 0.3786,
      "step": 251
    },
    {
      "epoch": 1.63,
      "learning_rate": 8.529964795802485e-05,
      "loss": 0.5154,
      "step": 252
    },
    {
      "epoch": 1.63,
      "learning_rate": 8.518800355015892e-05,
      "loss": 0.3841,
      "step": 253
    },
    {
      "epoch": 1.64,
      "learning_rate": 8.507601045756085e-05,
      "loss": 0.4477,
      "step": 254
    },
    {
      "epoch": 1.65,
      "learning_rate": 8.49636697899919e-05,
      "loss": 0.5142,
      "step": 255
    },
    {
      "epoch": 1.65,
      "learning_rate": 8.485098266065744e-05,
      "loss": 0.3414,
      "step": 256
    },
    {
      "epoch": 1.66,
      "learning_rate": 8.473795018619604e-05,
      "loss": 0.5165,
      "step": 257
    },
    {
      "epoch": 1.66,
      "learning_rate": 8.462457348666835e-05,
      "loss": 0.4612,
      "step": 258
    },
    {
      "epoch": 1.67,
      "learning_rate": 8.4510853685546e-05,
      "loss": 0.5629,
      "step": 259
    },
    {
      "epoch": 1.68,
      "learning_rate": 8.439679190970052e-05,
      "loss": 0.5442,
      "step": 260
    },
    {
      "epoch": 1.68,
      "learning_rate": 8.428238928939207e-05,
      "loss": 0.5181,
      "step": 261
    },
    {
      "epoch": 1.69,
      "learning_rate": 8.416764695825835e-05,
      "loss": 0.3429,
      "step": 262
    },
    {
      "epoch": 1.7,
      "learning_rate": 8.405256605330331e-05,
      "loss": 0.3904,
      "step": 263
    },
    {
      "epoch": 1.7,
      "learning_rate": 8.39371477148859e-05,
      "loss": 0.3796,
      "step": 264
    },
    {
      "epoch": 1.71,
      "learning_rate": 8.382139308670875e-05,
      "loss": 0.4801,
      "step": 265
    },
    {
      "epoch": 1.72,
      "learning_rate": 8.370530331580686e-05,
      "loss": 0.4761,
      "step": 266
    },
    {
      "epoch": 1.72,
      "learning_rate": 8.35888795525362e-05,
      "loss": 0.4367,
      "step": 267
    },
    {
      "epoch": 1.73,
      "learning_rate": 8.347212295056239e-05,
      "loss": 0.4171,
      "step": 268
    },
    {
      "epoch": 1.74,
      "learning_rate": 8.335503466684915e-05,
      "loss": 0.4233,
      "step": 269
    },
    {
      "epoch": 1.74,
      "learning_rate": 8.323761586164695e-05,
      "loss": 0.7066,
      "step": 270
    },
    {
      "epoch": 1.75,
      "learning_rate": 8.311986769848141e-05,
      "loss": 0.4359,
      "step": 271
    },
    {
      "epoch": 1.75,
      "learning_rate": 8.300179134414188e-05,
      "loss": 0.5577,
      "step": 272
    },
    {
      "epoch": 1.76,
      "learning_rate": 8.288338796866976e-05,
      "loss": 0.4351,
      "step": 273
    },
    {
      "epoch": 1.77,
      "learning_rate": 8.276465874534702e-05,
      "loss": 0.5204,
      "step": 274
    },
    {
      "epoch": 1.77,
      "learning_rate": 8.264560485068446e-05,
      "loss": 0.4543,
      "step": 275
    },
    {
      "epoch": 1.78,
      "learning_rate": 8.252622746441021e-05,
      "loss": 0.4326,
      "step": 276
    },
    {
      "epoch": 1.79,
      "learning_rate": 8.240652776945781e-05,
      "loss": 0.551,
      "step": 277
    },
    {
      "epoch": 1.79,
      "learning_rate": 8.228650695195472e-05,
      "loss": 0.5187,
      "step": 278
    },
    {
      "epoch": 1.8,
      "learning_rate": 8.216616620121043e-05,
      "loss": 0.4422,
      "step": 279
    },
    {
      "epoch": 1.81,
      "learning_rate": 8.204550670970469e-05,
      "loss": 0.6078,
      "step": 280
    },
    {
      "epoch": 1.81,
      "learning_rate": 8.192452967307576e-05,
      "loss": 0.5152,
      "step": 281
    },
    {
      "epoch": 1.82,
      "learning_rate": 8.180323629010848e-05,
      "loss": 0.5467,
      "step": 282
    },
    {
      "epoch": 1.83,
      "learning_rate": 8.168162776272244e-05,
      "loss": 0.6686,
      "step": 283
    },
    {
      "epoch": 1.83,
      "learning_rate": 8.155970529596006e-05,
      "loss": 0.44,
      "step": 284
    },
    {
      "epoch": 1.84,
      "learning_rate": 8.143747009797464e-05,
      "loss": 0.4946,
      "step": 285
    },
    {
      "epoch": 1.85,
      "learning_rate": 8.131492338001839e-05,
      "loss": 0.5184,
      "step": 286
    },
    {
      "epoch": 1.85,
      "learning_rate": 8.119206635643045e-05,
      "loss": 0.5919,
      "step": 287
    },
    {
      "epoch": 1.86,
      "learning_rate": 8.106890024462481e-05,
      "loss": 0.5617,
      "step": 288
    },
    {
      "epoch": 1.86,
      "learning_rate": 8.094542626507828e-05,
      "loss": 0.4162,
      "step": 289
    },
    {
      "epoch": 1.87,
      "learning_rate": 8.082164564131845e-05,
      "loss": 0.4794,
      "step": 290
    },
    {
      "epoch": 1.88,
      "learning_rate": 8.069755959991142e-05,
      "loss": 0.507,
      "step": 291
    },
    {
      "epoch": 1.88,
      "learning_rate": 8.057316937044977e-05,
      "loss": 0.5496,
      "step": 292
    },
    {
      "epoch": 1.89,
      "learning_rate": 8.044847618554034e-05,
      "loss": 0.5765,
      "step": 293
    },
    {
      "epoch": 1.9,
      "learning_rate": 8.032348128079203e-05,
      "loss": 0.5214,
      "step": 294
    },
    {
      "epoch": 1.9,
      "learning_rate": 8.019818589480352e-05,
      "loss": 0.3886,
      "step": 295
    },
    {
      "epoch": 1.91,
      "learning_rate": 8.0072591269151e-05,
      "loss": 0.3821,
      "step": 296
    },
    {
      "epoch": 1.92,
      "learning_rate": 7.994669864837594e-05,
      "loss": 0.5564,
      "step": 297
    },
    {
      "epoch": 1.92,
      "learning_rate": 7.982050927997264e-05,
      "loss": 0.6412,
      "step": 298
    },
    {
      "epoch": 1.93,
      "learning_rate": 7.969402441437594e-05,
      "loss": 0.5426,
      "step": 299
    },
    {
      "epoch": 1.94,
      "learning_rate": 7.956724530494887e-05,
      "loss": 0.538,
      "step": 300
    },
    {
      "epoch": 1.94,
      "learning_rate": 7.944017320797013e-05,
      "loss": 0.5215,
      "step": 301
    },
    {
      "epoch": 1.95,
      "learning_rate": 7.931280938262169e-05,
      "loss": 0.65,
      "step": 302
    },
    {
      "epoch": 1.95,
      "learning_rate": 7.918515509097634e-05,
      "loss": 0.6064,
      "step": 303
    },
    {
      "epoch": 1.96,
      "learning_rate": 7.905721159798513e-05,
      "loss": 0.4269,
      "step": 304
    },
    {
      "epoch": 1.97,
      "learning_rate": 7.89289801714649e-05,
      "loss": 0.5641,
      "step": 305
    },
    {
      "epoch": 1.97,
      "learning_rate": 7.880046208208563e-05,
      "loss": 0.4766,
      "step": 306
    },
    {
      "epoch": 1.98,
      "learning_rate": 7.867165860335792e-05,
      "loss": 0.3531,
      "step": 307
    },
    {
      "epoch": 1.99,
      "learning_rate": 7.854257101162037e-05,
      "loss": 0.5067,
      "step": 308
    },
    {
      "epoch": 1.99,
      "learning_rate": 7.841320058602688e-05,
      "loss": 0.4465,
      "step": 309
    },
    {
      "epoch": 2.0,
      "learning_rate": 7.828354860853399e-05,
      "loss": 0.8233,
      "step": 310
    },
    {
      "epoch": 2.01,
      "learning_rate": 7.815361636388827e-05,
      "loss": 0.3403,
      "step": 311
    },
    {
      "epoch": 2.01,
      "learning_rate": 7.802340513961342e-05,
      "loss": 0.4297,
      "step": 312
    },
    {
      "epoch": 2.02,
      "learning_rate": 7.789291622599767e-05,
      "loss": 0.3845,
      "step": 313
    },
    {
      "epoch": 2.03,
      "learning_rate": 7.776215091608085e-05,
      "loss": 0.3389,
      "step": 314
    },
    {
      "epoch": 2.03,
      "learning_rate": 7.763111050564178e-05,
      "loss": 0.377,
      "step": 315
    },
    {
      "epoch": 2.04,
      "learning_rate": 7.749979629318516e-05,
      "loss": 0.4526,
      "step": 316
    },
    {
      "epoch": 2.05,
      "learning_rate": 7.736820957992895e-05,
      "loss": 0.4133,
      "step": 317
    },
    {
      "epoch": 2.05,
      "learning_rate": 7.723635166979133e-05,
      "loss": 0.2625,
      "step": 318
    },
    {
      "epoch": 2.06,
      "learning_rate": 7.710422386937784e-05,
      "loss": 0.2808,
      "step": 319
    },
    {
      "epoch": 2.06,
      "learning_rate": 7.697182748796841e-05,
      "loss": 0.2915,
      "step": 320
    },
    {
      "epoch": 2.07,
      "learning_rate": 7.683916383750436e-05,
      "loss": 0.4015,
      "step": 321
    },
    {
      "epoch": 2.08,
      "learning_rate": 7.670623423257548e-05,
      "loss": 0.2656,
      "step": 322
    },
    {
      "epoch": 2.08,
      "learning_rate": 7.657303999040693e-05,
      "loss": 0.409,
      "step": 323
    },
    {
      "epoch": 2.09,
      "learning_rate": 7.64395824308462e-05,
      "loss": 0.4225,
      "step": 324
    },
    {
      "epoch": 2.1,
      "learning_rate": 7.630586287635008e-05,
      "loss": 0.3039,
      "step": 325
    },
    {
      "epoch": 2.1,
      "learning_rate": 7.617188265197148e-05,
      "loss": 0.3898,
      "step": 326
    },
    {
      "epoch": 2.11,
      "learning_rate": 7.603764308534636e-05,
      "loss": 0.331,
      "step": 327
    },
    {
      "epoch": 2.12,
      "learning_rate": 7.590314550668054e-05,
      "loss": 0.287,
      "step": 328
    },
    {
      "epoch": 2.12,
      "learning_rate": 7.576839124873653e-05,
      "loss": 0.3426,
      "step": 329
    },
    {
      "epoch": 2.13,
      "learning_rate": 7.563338164682036e-05,
      "loss": 0.3147,
      "step": 330
    },
    {
      "epoch": 2.14,
      "learning_rate": 7.549811803876825e-05,
      "loss": 0.4473,
      "step": 331
    },
    {
      "epoch": 2.14,
      "learning_rate": 7.536260176493348e-05,
      "loss": 0.3059,
      "step": 332
    },
    {
      "epoch": 2.15,
      "learning_rate": 7.5226834168173e-05,
      "loss": 0.2256,
      "step": 333
    },
    {
      "epoch": 2.15,
      "learning_rate": 7.509081659383417e-05,
      "loss": 0.2517,
      "step": 334
    },
    {
      "epoch": 2.16,
      "learning_rate": 7.495455038974146e-05,
      "loss": 0.3181,
      "step": 335
    },
    {
      "epoch": 2.17,
      "learning_rate": 7.481803690618304e-05,
      "loss": 0.3249,
      "step": 336
    },
    {
      "epoch": 2.17,
      "learning_rate": 7.46812774958974e-05,
      "loss": 0.3898,
      "step": 337
    },
    {
      "epoch": 2.18,
      "learning_rate": 7.454427351405999e-05,
      "loss": 0.4129,
      "step": 338
    },
    {
      "epoch": 2.19,
      "learning_rate": 7.440702631826977e-05,
      "loss": 0.3327,
      "step": 339
    },
    {
      "epoch": 2.19,
      "learning_rate": 7.426953726853574e-05,
      "loss": 0.2675,
      "step": 340
    },
    {
      "epoch": 2.2,
      "learning_rate": 7.413180772726348e-05,
      "loss": 0.2598,
      "step": 341
    },
    {
      "epoch": 2.21,
      "learning_rate": 7.399383905924165e-05,
      "loss": 0.2614,
      "step": 342
    },
    {
      "epoch": 2.21,
      "learning_rate": 7.385563263162847e-05,
      "loss": 0.483,
      "step": 343
    },
    {
      "epoch": 2.22,
      "learning_rate": 7.371718981393815e-05,
      "loss": 0.4278,
      "step": 344
    },
    {
      "epoch": 2.23,
      "learning_rate": 7.357851197802735e-05,
      "loss": 0.3948,
      "step": 345
    },
    {
      "epoch": 2.23,
      "learning_rate": 7.343960049808156e-05,
      "loss": 0.2719,
      "step": 346
    },
    {
      "epoch": 2.24,
      "learning_rate": 7.330045675060149e-05,
      "loss": 0.334,
      "step": 347
    },
    {
      "epoch": 2.25,
      "learning_rate": 7.316108211438945e-05,
      "loss": 0.2992,
      "step": 348
    },
    {
      "epoch": 2.25,
      "learning_rate": 7.302147797053569e-05,
      "loss": 0.3979,
      "step": 349
    },
    {
      "epoch": 2.26,
      "learning_rate": 7.288164570240463e-05,
      "loss": 0.2343,
      "step": 350
    },
    {
      "epoch": 2.26,
      "learning_rate": 7.274158669562126e-05,
      "loss": 0.3736,
      "step": 351
    },
    {
      "epoch": 2.27,
      "learning_rate": 7.26013023380574e-05,
      "loss": 0.3414,
      "step": 352
    },
    {
      "epoch": 2.28,
      "learning_rate": 7.246079401981784e-05,
      "loss": 0.3553,
      "step": 353
    },
    {
      "epoch": 2.28,
      "learning_rate": 7.232006313322667e-05,
      "loss": 0.2938,
      "step": 354
    },
    {
      "epoch": 2.29,
      "learning_rate": 7.217911107281352e-05,
      "loss": 0.3578,
      "step": 355
    },
    {
      "epoch": 2.3,
      "learning_rate": 7.203793923529956e-05,
      "loss": 0.2291,
      "step": 356
    },
    {
      "epoch": 2.3,
      "learning_rate": 7.189654901958385e-05,
      "loss": 0.3357,
      "step": 357
    },
    {
      "epoch": 2.31,
      "learning_rate": 7.175494182672939e-05,
      "loss": 0.2978,
      "step": 358
    },
    {
      "epoch": 2.32,
      "learning_rate": 7.161311905994922e-05,
      "loss": 0.324,
      "step": 359
    },
    {
      "epoch": 2.32,
      "learning_rate": 7.147108212459257e-05,
      "loss": 0.2886,
      "step": 360
    },
    {
      "epoch": 2.33,
      "learning_rate": 7.13288324281309e-05,
      "loss": 0.3199,
      "step": 361
    },
    {
      "epoch": 2.34,
      "learning_rate": 7.118637138014396e-05,
      "loss": 0.347,
      "step": 362
    },
    {
      "epoch": 2.34,
      "learning_rate": 7.104370039230583e-05,
      "loss": 0.3261,
      "step": 363
    },
    {
      "epoch": 2.35,
      "learning_rate": 7.090082087837092e-05,
      "loss": 0.2199,
      "step": 364
    },
    {
      "epoch": 2.35,
      "learning_rate": 7.075773425415994e-05,
      "loss": 0.3199,
      "step": 365
    },
    {
      "epoch": 2.36,
      "learning_rate": 7.061444193754596e-05,
      "loss": 0.4151,
      "step": 366
    },
    {
      "epoch": 2.37,
      "learning_rate": 7.047094534844023e-05,
      "loss": 0.3067,
      "step": 367
    },
    {
      "epoch": 2.37,
      "learning_rate": 7.032724590877821e-05,
      "loss": 0.3538,
      "step": 368
    },
    {
      "epoch": 2.38,
      "learning_rate": 7.018334504250545e-05,
      "loss": 0.2958,
      "step": 369
    },
    {
      "epoch": 2.39,
      "learning_rate": 7.003924417556343e-05,
      "loss": 0.2759,
      "step": 370
    },
    {
      "epoch": 2.39,
      "learning_rate": 6.989494473587554e-05,
      "loss": 0.2512,
      "step": 371
    },
    {
      "epoch": 2.4,
      "learning_rate": 6.975044815333282e-05,
      "loss": 0.2502,
      "step": 372
    },
    {
      "epoch": 2.41,
      "learning_rate": 6.960575585977984e-05,
      "loss": 0.2968,
      "step": 373
    },
    {
      "epoch": 2.41,
      "learning_rate": 6.946086928900054e-05,
      "loss": 0.2694,
      "step": 374
    },
    {
      "epoch": 2.42,
      "learning_rate": 6.931578987670396e-05,
      "loss": 0.3846,
      "step": 375
    },
    {
      "epoch": 2.43,
      "learning_rate": 6.917051906051006e-05,
      "loss": 0.3195,
      "step": 376
    },
    {
      "epoch": 2.43,
      "learning_rate": 6.902505827993541e-05,
      "loss": 0.2422,
      "step": 377
    },
    {
      "epoch": 2.44,
      "learning_rate": 6.887940897637908e-05,
      "loss": 0.3976,
      "step": 378
    },
    {
      "epoch": 2.45,
      "learning_rate": 6.873357259310815e-05,
      "loss": 0.3997,
      "step": 379
    },
    {
      "epoch": 2.45,
      "learning_rate": 6.858755057524354e-05,
      "loss": 0.3225,
      "step": 380
    },
    {
      "epoch": 2.46,
      "learning_rate": 6.844134436974567e-05,
      "loss": 0.356,
      "step": 381
    },
    {
      "epoch": 2.46,
      "learning_rate": 6.829495542540013e-05,
      "loss": 0.4587,
      "step": 382
    },
    {
      "epoch": 2.47,
      "learning_rate": 6.814838519280324e-05,
      "loss": 0.3709,
      "step": 383
    },
    {
      "epoch": 2.48,
      "learning_rate": 6.80016351243478e-05,
      "loss": 0.3335,
      "step": 384
    },
    {
      "epoch": 2.48,
      "learning_rate": 6.785470667420862e-05,
      "loss": 0.3348,
      "step": 385
    },
    {
      "epoch": 2.49,
      "learning_rate": 6.77076012983281e-05,
      "loss": 0.395,
      "step": 386
    },
    {
      "epoch": 2.5,
      "learning_rate": 6.75603204544019e-05,
      "loss": 0.3819,
      "step": 387
    },
    {
      "epoch": 2.5,
      "learning_rate": 6.741286560186437e-05,
      "loss": 0.3592,
      "step": 388
    },
    {
      "epoch": 2.51,
      "learning_rate": 6.726523820187413e-05,
      "loss": 0.3216,
      "step": 389
    },
    {
      "epoch": 2.52,
      "learning_rate": 6.711743971729967e-05,
      "loss": 0.3421,
      "step": 390
    },
    {
      "epoch": 2.52,
      "learning_rate": 6.696947161270476e-05,
      "loss": 0.2896,
      "step": 391
    },
    {
      "epoch": 2.53,
      "learning_rate": 6.682133535433393e-05,
      "loss": 0.3415,
      "step": 392
    },
    {
      "epoch": 2.54,
      "learning_rate": 6.667303241009803e-05,
      "loss": 0.2549,
      "step": 393
    },
    {
      "epoch": 2.54,
      "learning_rate": 6.652456424955963e-05,
      "loss": 0.2978,
      "step": 394
    },
    {
      "epoch": 2.55,
      "learning_rate": 6.637593234391843e-05,
      "loss": 0.234,
      "step": 395
    },
    {
      "epoch": 2.55,
      "learning_rate": 6.622713816599673e-05,
      "loss": 0.3472,
      "step": 396
    },
    {
      "epoch": 2.56,
      "learning_rate": 6.60781831902248e-05,
      "loss": 0.3334,
      "step": 397
    },
    {
      "epoch": 2.57,
      "learning_rate": 6.592906889262632e-05,
      "loss": 0.4193,
      "step": 398
    },
    {
      "epoch": 2.57,
      "learning_rate": 6.577979675080369e-05,
      "loss": 0.2992,
      "step": 399
    },
    {
      "epoch": 2.58,
      "learning_rate": 6.563036824392344e-05,
      "loss": 0.3312,
      "step": 400
    },
    {
      "epoch": 2.59,
      "learning_rate": 6.548078485270152e-05,
      "loss": 0.3205,
      "step": 401
    },
    {
      "epoch": 2.59,
      "learning_rate": 6.533104805938873e-05,
      "loss": 0.4147,
      "step": 402
    },
    {
      "epoch": 2.6,
      "learning_rate": 6.518115934775585e-05,
      "loss": 0.2941,
      "step": 403
    },
    {
      "epoch": 2.61,
      "learning_rate": 6.503112020307916e-05,
      "loss": 0.2874,
      "step": 404
    },
    {
      "epoch": 2.61,
      "learning_rate": 6.488093211212555e-05,
      "loss": 0.245,
      "step": 405
    },
    {
      "epoch": 2.62,
      "learning_rate": 6.473059656313782e-05,
      "loss": 0.3884,
      "step": 406
    },
    {
      "epoch": 2.63,
      "learning_rate": 6.458011504582005e-05,
      "loss": 0.3488,
      "step": 407
    },
    {
      "epoch": 2.63,
      "learning_rate": 6.442948905132266e-05,
      "loss": 0.3391,
      "step": 408
    },
    {
      "epoch": 2.64,
      "learning_rate": 6.427872007222777e-05,
      "loss": 0.2685,
      "step": 409
    },
    {
      "epoch": 2.65,
      "learning_rate": 6.412780960253436e-05,
      "loss": 0.3603,
      "step": 410
    },
    {
      "epoch": 2.65,
      "learning_rate": 6.397675913764347e-05,
      "loss": 0.3248,
      "step": 411
    },
    {
      "epoch": 2.66,
      "learning_rate": 6.382557017434332e-05,
      "loss": 0.3787,
      "step": 412
    },
    {
      "epoch": 2.66,
      "learning_rate": 6.367424421079463e-05,
      "loss": 0.3143,
      "step": 413
    },
    {
      "epoch": 2.67,
      "learning_rate": 6.352278274651561e-05,
      "loss": 0.2648,
      "step": 414
    },
    {
      "epoch": 2.68,
      "learning_rate": 6.337118728236721e-05,
      "loss": 0.2748,
      "step": 415
    },
    {
      "epoch": 2.68,
      "learning_rate": 6.321945932053822e-05,
      "loss": 0.2539,
      "step": 416
    },
    {
      "epoch": 2.69,
      "learning_rate": 6.306760036453035e-05,
      "loss": 0.2781,
      "step": 417
    },
    {
      "epoch": 2.7,
      "learning_rate": 6.291561191914333e-05,
      "loss": 0.2791,
      "step": 418
    },
    {
      "epoch": 2.7,
      "learning_rate": 6.276349549046007e-05,
      "loss": 0.3121,
      "step": 419
    },
    {
      "epoch": 2.71,
      "learning_rate": 6.261125258583171e-05,
      "loss": 0.2881,
      "step": 420
    },
    {
      "epoch": 2.72,
      "learning_rate": 6.245888471386263e-05,
      "loss": 0.4366,
      "step": 421
    },
    {
      "epoch": 2.72,
      "learning_rate": 6.230639338439549e-05,
      "loss": 0.4178,
      "step": 422
    },
    {
      "epoch": 2.73,
      "learning_rate": 6.215378010849641e-05,
      "loss": 0.305,
      "step": 423
    },
    {
      "epoch": 2.74,
      "learning_rate": 6.200104639843985e-05,
      "loss": 0.2393,
      "step": 424
    },
    {
      "epoch": 2.74,
      "learning_rate": 6.184819376769364e-05,
      "loss": 0.2826,
      "step": 425
    },
    {
      "epoch": 2.75,
      "learning_rate": 6.169522373090412e-05,
      "loss": 0.2871,
      "step": 426
    },
    {
      "epoch": 2.75,
      "learning_rate": 6.154213780388092e-05,
      "loss": 0.254,
      "step": 427
    },
    {
      "epoch": 2.76,
      "learning_rate": 6.138893750358212e-05,
      "loss": 0.3175,
      "step": 428
    },
    {
      "epoch": 2.77,
      "learning_rate": 6.123562434809912e-05,
      "loss": 0.3756,
      "step": 429
    },
    {
      "epoch": 2.77,
      "learning_rate": 6.108219985664161e-05,
      "loss": 0.301,
      "step": 430
    },
    {
      "epoch": 2.78,
      "learning_rate": 6.0928665549522554e-05,
      "loss": 0.4171,
      "step": 431
    },
    {
      "epoch": 2.79,
      "learning_rate": 6.0775022948143115e-05,
      "loss": 0.2624,
      "step": 432
    },
    {
      "epoch": 2.79,
      "learning_rate": 6.06212735749775e-05,
      "loss": 0.3901,
      "step": 433
    },
    {
      "epoch": 2.8,
      "learning_rate": 6.046741895355802e-05,
      "loss": 0.3972,
      "step": 434
    },
    {
      "epoch": 2.81,
      "learning_rate": 6.031346060845986e-05,
      "loss": 0.2281,
      "step": 435
    },
    {
      "epoch": 2.81,
      "learning_rate": 6.015940006528602e-05,
      "loss": 0.2912,
      "step": 436
    },
    {
      "epoch": 2.82,
      "learning_rate": 6.0005238850652234e-05,
      "loss": 0.3097,
      "step": 437
    },
    {
      "epoch": 2.83,
      "learning_rate": 5.9850978492171794e-05,
      "loss": 0.2776,
      "step": 438
    },
    {
      "epoch": 2.83,
      "learning_rate": 5.96966205184404e-05,
      "loss": 0.366,
      "step": 439
    },
    {
      "epoch": 2.84,
      "learning_rate": 5.954216645902109e-05,
      "loss": 0.2972,
      "step": 440
    },
    {
      "epoch": 2.85,
      "learning_rate": 5.9387617844429e-05,
      "loss": 0.2476,
      "step": 441
    },
    {
      "epoch": 2.85,
      "learning_rate": 5.923297620611623e-05,
      "loss": 0.2575,
      "step": 442
    },
    {
      "epoch": 2.86,
      "learning_rate": 5.907824307645669e-05,
      "loss": 0.3426,
      "step": 443
    },
    {
      "epoch": 2.86,
      "learning_rate": 5.892341998873089e-05,
      "loss": 0.2804,
      "step": 444
    },
    {
      "epoch": 2.87,
      "learning_rate": 5.876850847711073e-05,
      "loss": 0.4664,
      "step": 445
    },
    {
      "epoch": 2.88,
      "learning_rate": 5.861351007664434e-05,
      "loss": 0.2552,
      "step": 446
    },
    {
      "epoch": 2.88,
      "learning_rate": 5.845842632324088e-05,
      "loss": 0.2967,
      "step": 447
    },
    {
      "epoch": 2.89,
      "learning_rate": 5.83032587536552e-05,
      "loss": 0.3264,
      "step": 448
    },
    {
      "epoch": 2.9,
      "learning_rate": 5.814800890547278e-05,
      "loss": 0.4055,
      "step": 449
    },
    {
      "epoch": 2.9,
      "learning_rate": 5.799267831709442e-05,
      "loss": 0.293,
      "step": 450
    },
    {
      "epoch": 2.91,
      "learning_rate": 5.78372685277209e-05,
      "loss": 0.4247,
      "step": 451
    },
    {
      "epoch": 2.92,
      "learning_rate": 5.7681781077337905e-05,
      "loss": 0.3368,
      "step": 452
    },
    {
      "epoch": 2.92,
      "learning_rate": 5.752621750670068e-05,
      "loss": 0.3454,
      "step": 453
    },
    {
      "epoch": 2.93,
      "learning_rate": 5.737057935731868e-05,
      "loss": 0.3367,
      "step": 454
    },
    {
      "epoch": 2.94,
      "learning_rate": 5.721486817144044e-05,
      "loss": 0.3756,
      "step": 455
    },
    {
      "epoch": 2.94,
      "learning_rate": 5.705908549203823e-05,
      "loss": 0.47,
      "step": 456
    },
    {
      "epoch": 2.95,
      "learning_rate": 5.690323286279274e-05,
      "loss": 0.3024,
      "step": 457
    },
    {
      "epoch": 2.95,
      "learning_rate": 5.674731182807781e-05,
      "loss": 0.2348,
      "step": 458
    },
    {
      "epoch": 2.96,
      "learning_rate": 5.659132393294514e-05,
      "loss": 0.4405,
      "step": 459
    },
    {
      "epoch": 2.97,
      "learning_rate": 5.643527072310891e-05,
      "loss": 0.3661,
      "step": 460
    },
    {
      "epoch": 2.97,
      "learning_rate": 5.627915374493061e-05,
      "loss": 0.3194,
      "step": 461
    },
    {
      "epoch": 2.98,
      "learning_rate": 5.612297454540352e-05,
      "loss": 0.3614,
      "step": 462
    },
    {
      "epoch": 2.99,
      "learning_rate": 5.596673467213756e-05,
      "loss": 0.3361,
      "step": 463
    },
    {
      "epoch": 2.99,
      "learning_rate": 5.581043567334383e-05,
      "loss": 0.2992,
      "step": 464
    },
    {
      "epoch": 3.0,
      "learning_rate": 5.5654079097819345e-05,
      "loss": 0.2454,
      "step": 465
    },
    {
      "epoch": 3.01,
      "learning_rate": 5.5497666494931654e-05,
      "loss": 0.1756,
      "step": 466
    },
    {
      "epoch": 3.01,
      "learning_rate": 5.5341199414603493e-05,
      "loss": 0.1941,
      "step": 467
    },
    {
      "epoch": 3.02,
      "learning_rate": 5.518467940729739e-05,
      "loss": 0.1756,
      "step": 468
    },
    {
      "epoch": 3.03,
      "learning_rate": 5.502810802400039e-05,
      "loss": 0.1196,
      "step": 469
    },
    {
      "epoch": 3.03,
      "learning_rate": 5.487148681620862e-05,
      "loss": 0.1895,
      "step": 470
    },
    {
      "epoch": 3.04,
      "learning_rate": 5.4714817335911894e-05,
      "loss": 0.2111,
      "step": 471
    },
    {
      "epoch": 3.05,
      "learning_rate": 5.455810113557839e-05,
      "loss": 0.1684,
      "step": 472
    },
    {
      "epoch": 3.05,
      "learning_rate": 5.440133976813926e-05,
      "loss": 0.1822,
      "step": 473
    },
    {
      "epoch": 3.06,
      "learning_rate": 5.4244534786973214e-05,
      "loss": 0.1987,
      "step": 474
    },
    {
      "epoch": 3.06,
      "learning_rate": 5.40876877458911e-05,
      "loss": 0.1747,
      "step": 475
    },
    {
      "epoch": 3.07,
      "learning_rate": 5.3930800199120616e-05,
      "loss": 0.2005,
      "step": 476
    },
    {
      "epoch": 3.08,
      "learning_rate": 5.377387370129079e-05,
      "loss": 0.1414,
      "step": 477
    },
    {
      "epoch": 3.08,
      "learning_rate": 5.361690980741663e-05,
      "loss": 0.1509,
      "step": 478
    },
    {
      "epoch": 3.09,
      "learning_rate": 5.345991007288371e-05,
      "loss": 0.2247,
      "step": 479
    },
    {
      "epoch": 3.1,
      "learning_rate": 5.330287605343279e-05,
      "loss": 0.1593,
      "step": 480
    },
    {
      "epoch": 3.1,
      "learning_rate": 5.314580930514431e-05,
      "loss": 0.1525,
      "step": 481
    },
    {
      "epoch": 3.11,
      "learning_rate": 5.298871138442307e-05,
      "loss": 0.1869,
      "step": 482
    },
    {
      "epoch": 3.12,
      "learning_rate": 5.283158384798275e-05,
      "loss": 0.1962,
      "step": 483
    },
    {
      "epoch": 3.12,
      "learning_rate": 5.267442825283048e-05,
      "loss": 0.2473,
      "step": 484
    },
    {
      "epoch": 3.13,
      "learning_rate": 5.2517246156251455e-05,
      "loss": 0.1982,
      "step": 485
    },
    {
      "epoch": 3.14,
      "learning_rate": 5.236003911579345e-05,
      "loss": 0.1662,
      "step": 486
    },
    {
      "epoch": 3.14,
      "learning_rate": 5.220280868925145e-05,
      "loss": 0.1432,
      "step": 487
    },
    {
      "epoch": 3.15,
      "learning_rate": 5.204555643465215e-05,
      "loss": 0.1959,
      "step": 488
    },
    {
      "epoch": 3.15,
      "learning_rate": 5.1888283910238555e-05,
      "loss": 0.2394,
      "step": 489
    },
    {
      "epoch": 3.16,
      "learning_rate": 5.173099267445451e-05,
      "loss": 0.2186,
      "step": 490
    },
    {
      "epoch": 3.17,
      "learning_rate": 5.157368428592933e-05,
      "loss": 0.131,
      "step": 491
    },
    {
      "epoch": 3.17,
      "learning_rate": 5.1416360303462206e-05,
      "loss": 0.2114,
      "step": 492
    },
    {
      "epoch": 3.18,
      "learning_rate": 5.125902228600693e-05,
      "loss": 0.149,
      "step": 493
    },
    {
      "epoch": 3.19,
      "learning_rate": 5.110167179265636e-05,
      "loss": 0.1519,
      "step": 494
    },
    {
      "epoch": 3.19,
      "learning_rate": 5.094431038262693e-05,
      "loss": 0.1843,
      "step": 495
    },
    {
      "epoch": 3.2,
      "learning_rate": 5.078693961524329e-05,
      "loss": 0.2178,
      "step": 496
    },
    {
      "epoch": 3.21,
      "learning_rate": 5.062956104992285e-05,
      "loss": 0.21,
      "step": 497
    },
    {
      "epoch": 3.21,
      "learning_rate": 5.0472176246160184e-05,
      "loss": 0.1293,
      "step": 498
    },
    {
      "epoch": 3.22,
      "learning_rate": 5.031478676351179e-05,
      "loss": 0.1549,
      "step": 499
    },
    {
      "epoch": 3.23,
      "learning_rate": 5.01573941615805e-05,
      "loss": 0.182,
      "step": 500
    }
  ],
  "max_steps": 1000,
  "num_train_epochs": 7,
  "total_flos": 3.648778025081242e+16,
  "trial_name": null,
  "trial_params": null
}