diff --git "a/checkpoint-1000/trainer_state.json" "b/checkpoint-1000/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-1000/trainer_state.json"
@@ -0,0 +1,6016 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 12.658227848101266,
+  "global_step": 1000,
+  "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
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 5e-05,
+      "loss": 0.0133,
+      "step": 501
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 4.984260583841953e-05,
+      "loss": 0.0237,
+      "step": 502
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 4.9685213236488216e-05,
+      "loss": 0.0089,
+      "step": 503
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 4.9527823753839834e-05,
+      "loss": 0.0106,
+      "step": 504
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 4.937043895007717e-05,
+      "loss": 0.008,
+      "step": 505
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 4.9213060384756716e-05,
+      "loss": 0.0117,
+      "step": 506
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 4.9055689617373084e-05,
+      "loss": 0.0253,
+      "step": 507
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 4.8898328207343666e-05,
+      "loss": 0.0134,
+      "step": 508
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 4.874097771399308e-05,
+      "loss": 0.0133,
+      "step": 509
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 4.858363969653781e-05,
+      "loss": 0.0103,
+      "step": 510
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 4.8426315714070684e-05,
+      "loss": 0.0247,
+      "step": 511
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 4.8269007325545506e-05,
+      "loss": 0.0187,
+      "step": 512
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 4.8111716089761456e-05,
+      "loss": 0.0308,
+      "step": 513
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 4.7954443565347865e-05,
+      "loss": 0.0157,
+      "step": 514
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 4.779719131074857e-05,
+      "loss": 0.0131,
+      "step": 515
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 4.7639960884206576e-05,
+      "loss": 0.019,
+      "step": 516
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 4.7482753843748564e-05,
+      "loss": 0.0252,
+      "step": 517
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 4.7325571747169545e-05,
+      "loss": 0.0105,
+      "step": 518
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 4.716841615201726e-05,
+      "loss": 0.0082,
+      "step": 519
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 4.7011288615576934e-05,
+      "loss": 0.0233,
+      "step": 520
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 4.6854190694855694e-05,
+      "loss": 0.006,
+      "step": 521
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 4.6697123946567227e-05,
+      "loss": 0.0242,
+      "step": 522
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 4.65400899271163e-05,
+      "loss": 0.0136,
+      "step": 523
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 4.63830901925834e-05,
+      "loss": 0.0145,
+      "step": 524
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 4.6226126298709224e-05,
+      "loss": 0.0194,
+      "step": 525
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 4.60691998008794e-05,
+      "loss": 0.0284,
+      "step": 526
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 4.5912312254108905e-05,
+      "loss": 0.0179,
+      "step": 527
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 4.575546521302681e-05,
+      "loss": 0.0099,
+      "step": 528
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 4.5598660231860746e-05,
+      "loss": 0.0278,
+      "step": 529
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 4.544189886442162e-05,
+      "loss": 0.0177,
+      "step": 530
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 4.528518266408811e-05,
+      "loss": 0.0135,
+      "step": 531
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 4.5128513183791386e-05,
+      "loss": 0.0151,
+      "step": 532
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 4.49718919759996e-05,
+      "loss": 0.011,
+      "step": 533
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 4.481532059270262e-05,
+      "loss": 0.019,
+      "step": 534
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 4.465880058539652e-05,
+      "loss": 0.0063,
+      "step": 535
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 4.450233350506836e-05,
+      "loss": 0.0142,
+      "step": 536
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 4.4345920902180647e-05,
+      "loss": 0.0207,
+      "step": 537
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 4.418956432665618e-05,
+      "loss": 0.0142,
+      "step": 538
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 4.403326532786245e-05,
+      "loss": 0.0264,
+      "step": 539
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 4.387702545459649e-05,
+      "loss": 0.0193,
+      "step": 540
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 4.3720846255069406e-05,
+      "loss": 0.013,
+      "step": 541
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 4.356472927689109e-05,
+      "loss": 0.0275,
+      "step": 542
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 4.3408676067054866e-05,
+      "loss": 0.0122,
+      "step": 543
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 4.32526881719222e-05,
+      "loss": 0.0094,
+      "step": 544
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 4.3096767137207256e-05,
+      "loss": 0.0101,
+      "step": 545
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 4.2940914507961775e-05,
+      "loss": 0.026,
+      "step": 546
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 4.278513182855956e-05,
+      "loss": 0.0283,
+      "step": 547
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 4.262942064268134e-05,
+      "loss": 0.0126,
+      "step": 548
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 4.247378249329933e-05,
+      "loss": 0.0089,
+      "step": 549
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 4.23182189226621e-05,
+      "loss": 0.0065,
+      "step": 550
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 4.21627314722791e-05,
+      "loss": 0.0121,
+      "step": 551
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 4.20073216829056e-05,
+      "loss": 0.0219,
+      "step": 552
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 4.185199109452721e-05,
+      "loss": 0.0093,
+      "step": 553
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 4.169674124634481e-05,
+      "loss": 0.0086,
+      "step": 554
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 4.1541573676759126e-05,
+      "loss": 0.0115,
+      "step": 555
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 4.138648992335566e-05,
+      "loss": 0.0077,
+      "step": 556
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 4.12314915228893e-05,
+      "loss": 0.0094,
+      "step": 557
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 4.107658001126913e-05,
+      "loss": 0.0111,
+      "step": 558
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 4.092175692354333e-05,
+      "loss": 0.0113,
+      "step": 559
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 4.0767023793883785e-05,
+      "loss": 0.0049,
+      "step": 560
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 4.0612382155571026e-05,
+      "loss": 0.0119,
+      "step": 561
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 4.045783354097893e-05,
+      "loss": 0.0054,
+      "step": 562
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 4.0303379481559623e-05,
+      "loss": 0.0044,
+      "step": 563
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 4.0149021507828224e-05,
+      "loss": 0.015,
+      "step": 564
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 3.9994761149347784e-05,
+      "loss": 0.0097,
+      "step": 565
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 3.984059993471399e-05,
+      "loss": 0.005,
+      "step": 566
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 3.968653939154016e-05,
+      "loss": 0.0232,
+      "step": 567
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 3.9532581046442e-05,
+      "loss": 0.0282,
+      "step": 568
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 3.937872642502252e-05,
+      "loss": 0.0072,
+      "step": 569
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 3.9224977051856904e-05,
+      "loss": 0.0071,
+      "step": 570
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 3.907133445047747e-05,
+      "loss": 0.0049,
+      "step": 571
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 3.8917800143358404e-05,
+      "loss": 0.0107,
+      "step": 572
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 3.8764375651900906e-05,
+      "loss": 0.0043,
+      "step": 573
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 3.861106249641789e-05,
+      "loss": 0.0035,
+      "step": 574
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 3.84578621961191e-05,
+      "loss": 0.0046,
+      "step": 575
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 3.830477626909589e-05,
+      "loss": 0.007,
+      "step": 576
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 3.8151806232306374e-05,
+      "loss": 0.0116,
+      "step": 577
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 3.7998953601560175e-05,
+      "loss": 0.0069,
+      "step": 578
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 3.784621989150361e-05,
+      "loss": 0.0038,
+      "step": 579
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 3.769360661560453e-05,
+      "loss": 0.0138,
+      "step": 580
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 3.75411152861374e-05,
+      "loss": 0.0063,
+      "step": 581
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 3.73887474141683e-05,
+      "loss": 0.012,
+      "step": 582
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 3.723650450953994e-05,
+      "loss": 0.0058,
+      "step": 583
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 3.708438808085668e-05,
+      "loss": 0.0084,
+      "step": 584
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 3.693239963546967e-05,
+      "loss": 0.0049,
+      "step": 585
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 3.6780540679461784e-05,
+      "loss": 0.0054,
+      "step": 586
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 3.662881271763279e-05,
+      "loss": 0.0116,
+      "step": 587
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 3.64772172534844e-05,
+      "loss": 0.0077,
+      "step": 588
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 3.63257557892054e-05,
+      "loss": 0.0122,
+      "step": 589
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 3.6174429825656685e-05,
+      "loss": 0.0052,
+      "step": 590
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 3.602324086235655e-05,
+      "loss": 0.0046,
+      "step": 591
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 3.587219039746564e-05,
+      "loss": 0.0177,
+      "step": 592
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 3.572127992777223e-05,
+      "loss": 0.0059,
+      "step": 593
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 3.557051094867735e-05,
+      "loss": 0.0073,
+      "step": 594
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 3.541988495417997e-05,
+      "loss": 0.0054,
+      "step": 595
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 3.5269403436862175e-05,
+      "loss": 0.0046,
+      "step": 596
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 3.511906788787447e-05,
+      "loss": 0.0091,
+      "step": 597
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 3.496887979692084e-05,
+      "loss": 0.0124,
+      "step": 598
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 3.481884065224415e-05,
+      "loss": 0.01,
+      "step": 599
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 3.466895194061128e-05,
+      "loss": 0.0093,
+      "step": 600
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 3.451921514729848e-05,
+      "loss": 0.0074,
+      "step": 601
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 3.436963175607656e-05,
+      "loss": 0.0082,
+      "step": 602
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 3.422020324919632e-05,
+      "loss": 0.0042,
+      "step": 603
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 3.4070931107373675e-05,
+      "loss": 0.0079,
+      "step": 604
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 3.39218168097752e-05,
+      "loss": 0.0179,
+      "step": 605
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 3.377286183400328e-05,
+      "loss": 0.0151,
+      "step": 606
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 3.362406765608158e-05,
+      "loss": 0.0098,
+      "step": 607
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 3.3475435750440356e-05,
+      "loss": 0.0067,
+      "step": 608
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 3.332696758990197e-05,
+      "loss": 0.0094,
+      "step": 609
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 3.3178664645666066e-05,
+      "loss": 0.009,
+      "step": 610
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 3.303052838729525e-05,
+      "loss": 0.0036,
+      "step": 611
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 3.2882560282700336e-05,
+      "loss": 0.0071,
+      "step": 612
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 3.273476179812588e-05,
+      "loss": 0.0136,
+      "step": 613
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 3.258713439813566e-05,
+      "loss": 0.0078,
+      "step": 614
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 3.243967954559811e-05,
+      "loss": 0.0052,
+      "step": 615
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 3.229239870167191e-05,
+      "loss": 0.005,
+      "step": 616
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 3.2145293325791395e-05,
+      "loss": 0.0136,
+      "step": 617
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 3.199836487565222e-05,
+      "loss": 0.012,
+      "step": 618
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 3.1851614807196774e-05,
+      "loss": 0.0124,
+      "step": 619
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 3.170504457459989e-05,
+      "loss": 0.0115,
+      "step": 620
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 3.155865563025433e-05,
+      "loss": 0.0162,
+      "step": 621
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 3.1412449424756474e-05,
+      "loss": 0.0144,
+      "step": 622
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 3.1266427406891856e-05,
+      "loss": 0.0035,
+      "step": 623
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 3.112059102362093e-05,
+      "loss": 0.0089,
+      "step": 624
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 3.0974941720064585e-05,
+      "loss": 0.0086,
+      "step": 625
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 3.082948093948997e-05,
+      "loss": 0.0048,
+      "step": 626
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 3.0684210123296055e-05,
+      "loss": 0.0103,
+      "step": 627
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 3.053913071099947e-05,
+      "loss": 0.0196,
+      "step": 628
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 3.0394244140220163e-05,
+      "loss": 0.0043,
+      "step": 629
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 3.0249551846667207e-05,
+      "loss": 0.0074,
+      "step": 630
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 3.010505526412447e-05,
+      "loss": 0.0129,
+      "step": 631
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 2.996075582443658e-05,
+      "loss": 0.0035,
+      "step": 632
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 2.981665495749457e-05,
+      "loss": 0.0053,
+      "step": 633
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 2.9672754091221805e-05,
+      "loss": 0.0079,
+      "step": 634
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 2.9529054651559772e-05,
+      "loss": 0.0029,
+      "step": 635
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 2.938555806245406e-05,
+      "loss": 0.0061,
+      "step": 636
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 2.9242265745840063e-05,
+      "loss": 0.0077,
+      "step": 637
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 2.9099179121629117e-05,
+      "loss": 0.0035,
+      "step": 638
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 2.895629960769417e-05,
+      "loss": 0.0027,
+      "step": 639
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 2.881362861985606e-05,
+      "loss": 0.0063,
+      "step": 640
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 2.867116757186911e-05,
+      "loss": 0.0135,
+      "step": 641
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 2.8528917875407433e-05,
+      "loss": 0.0049,
+      "step": 642
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 2.838688094005078e-05,
+      "loss": 0.0027,
+      "step": 643
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 2.8245058173270622e-05,
+      "loss": 0.0089,
+      "step": 644
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 2.8103450980416136e-05,
+      "loss": 0.0056,
+      "step": 645
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 2.796206076470044e-05,
+      "loss": 0.0072,
+      "step": 646
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 2.7820888927186483e-05,
+      "loss": 0.0089,
+      "step": 647
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 2.7679936866773315e-05,
+      "loss": 0.0029,
+      "step": 648
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 2.753920598018217e-05,
+      "loss": 0.0045,
+      "step": 649
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 2.739869766194263e-05,
+      "loss": 0.0029,
+      "step": 650
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 2.7258413304378734e-05,
+      "loss": 0.0059,
+      "step": 651
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 2.7118354297595396e-05,
+      "loss": 0.012,
+      "step": 652
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 2.6978522029464325e-05,
+      "loss": 0.0115,
+      "step": 653
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 2.683891788561055e-05,
+      "loss": 0.0038,
+      "step": 654
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 2.669954324939852e-05,
+      "loss": 0.0068,
+      "step": 655
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 2.6560399501918465e-05,
+      "loss": 0.0059,
+      "step": 656
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 2.6421488021972673e-05,
+      "loss": 0.0072,
+      "step": 657
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 2.6282810186061862e-05,
+      "loss": 0.0039,
+      "step": 658
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 2.6144367368371535e-05,
+      "loss": 0.0029,
+      "step": 659
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 2.600616094075835e-05,
+      "loss": 0.0031,
+      "step": 660
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 2.5868192272736514e-05,
+      "loss": 0.0037,
+      "step": 661
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 2.5730462731464273e-05,
+      "loss": 0.006,
+      "step": 662
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 2.5592973681730236e-05,
+      "loss": 0.0029,
+      "step": 663
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 2.5455726485940012e-05,
+      "loss": 0.004,
+      "step": 664
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 2.5318722504102604e-05,
+      "loss": 0.0034,
+      "step": 665
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 2.5181963093816962e-05,
+      "loss": 0.0049,
+      "step": 666
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 2.504544961025853e-05,
+      "loss": 0.0053,
+      "step": 667
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 2.4909183406165836e-05,
+      "loss": 0.0113,
+      "step": 668
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 2.4773165831827018e-05,
+      "loss": 0.0086,
+      "step": 669
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 2.4637398235066527e-05,
+      "loss": 0.0053,
+      "step": 670
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 2.450188196123177e-05,
+      "loss": 0.0029,
+      "step": 671
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 2.4366618353179644e-05,
+      "loss": 0.0119,
+      "step": 672
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 2.423160875126348e-05,
+      "loss": 0.0025,
+      "step": 673
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 2.4096854493319477e-05,
+      "loss": 0.0021,
+      "step": 674
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 2.3962356914653657e-05,
+      "loss": 0.0033,
+      "step": 675
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 2.3828117348028528e-05,
+      "loss": 0.0037,
+      "step": 676
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 2.3694137123649946e-05,
+      "loss": 0.0025,
+      "step": 677
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 2.3560417569153796e-05,
+      "loss": 0.0036,
+      "step": 678
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 2.342696000959309e-05,
+      "loss": 0.003,
+      "step": 679
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 2.3293765767424537e-05,
+      "loss": 0.0023,
+      "step": 680
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 2.3160836162495653e-05,
+      "loss": 0.0023,
+      "step": 681
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 2.3028172512031604e-05,
+      "loss": 0.0122,
+      "step": 682
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 2.289577613062218e-05,
+      "loss": 0.0083,
+      "step": 683
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 2.276364833020868e-05,
+      "loss": 0.0059,
+      "step": 684
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 2.2631790420071064e-05,
+      "loss": 0.0028,
+      "step": 685
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 2.2500203706814856e-05,
+      "loss": 0.006,
+      "step": 686
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 2.2368889494358235e-05,
+      "loss": 0.0103,
+      "step": 687
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 2.2237849083919142e-05,
+      "loss": 0.0043,
+      "step": 688
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 2.2107083774002364e-05,
+      "loss": 0.0138,
+      "step": 689
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 2.1976594860386597e-05,
+      "loss": 0.0134,
+      "step": 690
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 2.1846383636111743e-05,
+      "loss": 0.0069,
+      "step": 691
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 2.1716451391466008e-05,
+      "loss": 0.0024,
+      "step": 692
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 2.1586799413973135e-05,
+      "loss": 0.0033,
+      "step": 693
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 2.1457428988379635e-05,
+      "loss": 0.0033,
+      "step": 694
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 2.1328341396642093e-05,
+      "loss": 0.0116,
+      "step": 695
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 2.1199537917914386e-05,
+      "loss": 0.0071,
+      "step": 696
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 2.107101982853511e-05,
+      "loss": 0.0067,
+      "step": 697
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 2.0942788402014867e-05,
+      "loss": 0.008,
+      "step": 698
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 2.0814844909023663e-05,
+      "loss": 0.0059,
+      "step": 699
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 2.068719061737831e-05,
+      "loss": 0.0054,
+      "step": 700
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 2.0559826792029884e-05,
+      "loss": 0.0021,
+      "step": 701
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 2.0432754695051136e-05,
+      "loss": 0.0058,
+      "step": 702
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 2.0305975585624058e-05,
+      "loss": 0.0081,
+      "step": 703
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 2.0179490720027372e-05,
+      "loss": 0.0051,
+      "step": 704
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 2.005330135162408e-05,
+      "loss": 0.0056,
+      "step": 705
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 1.992740873084899e-05,
+      "loss": 0.0045,
+      "step": 706
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 1.9801814105196497e-05,
+      "loss": 0.0043,
+      "step": 707
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 1.9676518719207977e-05,
+      "loss": 0.0051,
+      "step": 708
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 1.9551523814459665e-05,
+      "loss": 0.0038,
+      "step": 709
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 1.9426830629550242e-05,
+      "loss": 0.0052,
+      "step": 710
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 1.9302440400088606e-05,
+      "loss": 0.0167,
+      "step": 711
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 1.917835435868155e-05,
+      "loss": 0.0058,
+      "step": 712
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 1.9054573734921714e-05,
+      "loss": 0.0025,
+      "step": 713
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 1.8931099755375203e-05,
+      "loss": 0.0059,
+      "step": 714
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 1.880793364356956e-05,
+      "loss": 0.0063,
+      "step": 715
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 1.8685076619981608e-05,
+      "loss": 0.0034,
+      "step": 716
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 1.8562529902025372e-05,
+      "loss": 0.0048,
+      "step": 717
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 1.844029470403993e-05,
+      "loss": 0.0059,
+      "step": 718
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 1.8318372237277565e-05,
+      "loss": 0.003,
+      "step": 719
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 1.8196763709891524e-05,
+      "loss": 0.0019,
+      "step": 720
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 1.8075470326924243e-05,
+      "loss": 0.0024,
+      "step": 721
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 1.795449329029531e-05,
+      "loss": 0.0123,
+      "step": 722
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 1.7833833798789595e-05,
+      "loss": 0.0025,
+      "step": 723
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 1.7713493048045294e-05,
+      "loss": 0.0057,
+      "step": 724
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 1.7593472230542202e-05,
+      "loss": 0.0083,
+      "step": 725
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 1.747377253558982e-05,
+      "loss": 0.002,
+      "step": 726
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 1.7354395149315534e-05,
+      "loss": 0.0057,
+      "step": 727
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 1.7235341254653005e-05,
+      "loss": 0.0019,
+      "step": 728
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 1.7116612031330252e-05,
+      "loss": 0.0018,
+      "step": 729
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 1.6998208655858137e-05,
+      "loss": 0.0047,
+      "step": 730
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 1.6880132301518598e-05,
+      "loss": 0.004,
+      "step": 731
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 1.6762384138353078e-05,
+      "loss": 0.0034,
+      "step": 732
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 1.6644965333150847e-05,
+      "loss": 0.0073,
+      "step": 733
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 1.6527877049437622e-05,
+      "loss": 0.0155,
+      "step": 734
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 1.6411120447463807e-05,
+      "loss": 0.0086,
+      "step": 735
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 1.6294696684193154e-05,
+      "loss": 0.0108,
+      "step": 736
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 1.617860691329126e-05,
+      "loss": 0.0112,
+      "step": 737
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 1.6062852285114123e-05,
+      "loss": 0.0023,
+      "step": 738
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 1.5947433946696693e-05,
+      "loss": 0.002,
+      "step": 739
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 1.583235304174167e-05,
+      "loss": 0.0025,
+      "step": 740
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 1.5717610710607948e-05,
+      "loss": 0.004,
+      "step": 741
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 1.5603208090299498e-05,
+      "loss": 0.004,
+      "step": 742
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 1.5489146314454002e-05,
+      "loss": 0.002,
+      "step": 743
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 1.537542651333167e-05,
+      "loss": 0.0019,
+      "step": 744
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 1.5262049813803958e-05,
+      "loss": 0.009,
+      "step": 745
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 1.5149017339342574e-05,
+      "loss": 0.0113,
+      "step": 746
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 1.5036330210008115e-05,
+      "loss": 0.0025,
+      "step": 747
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 1.4923989542439159e-05,
+      "loss": 0.0044,
+      "step": 748
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 1.4811996449841098e-05,
+      "loss": 0.0029,
+      "step": 749
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 1.4700352041975168e-05,
+      "loss": 0.0023,
+      "step": 750
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 1.458905742514734e-05,
+      "loss": 0.0046,
+      "step": 751
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 1.447811370219757e-05,
+      "loss": 0.0028,
+      "step": 752
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 1.4367521972488612e-05,
+      "loss": 0.0031,
+      "step": 753
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 1.4257283331895315e-05,
+      "loss": 0.0037,
+      "step": 754
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 1.4147398872793693e-05,
+      "loss": 0.0023,
+      "step": 755
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 1.4037869684050115e-05,
+      "loss": 0.0018,
+      "step": 756
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 1.3928696851010443e-05,
+      "loss": 0.0048,
+      "step": 757
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 1.3819881455489458e-05,
+      "loss": 0.0101,
+      "step": 758
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 1.3711424575759912e-05,
+      "loss": 0.003,
+      "step": 759
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 1.3603327286542023e-05,
+      "loss": 0.0023,
+      "step": 760
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 1.3495590658992718e-05,
+      "loss": 0.0029,
+      "step": 761
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 1.33882157606951e-05,
+      "loss": 0.0042,
+      "step": 762
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 1.3281203655647756e-05,
+      "loss": 0.0059,
+      "step": 763
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 1.317455540425439e-05,
+      "loss": 0.0021,
+      "step": 764
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 1.3068272063313102e-05,
+      "loss": 0.0034,
+      "step": 765
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 1.2962354686006084e-05,
+      "loss": 0.013,
+      "step": 766
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 1.2856804321889115e-05,
+      "loss": 0.0031,
+      "step": 767
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 1.2751622016881182e-05,
+      "loss": 0.0028,
+      "step": 768
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 1.2646808813254035e-05,
+      "loss": 0.0061,
+      "step": 769
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 1.2542365749622049e-05,
+      "loss": 0.0037,
+      "step": 770
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 1.2438293860931677e-05,
+      "loss": 0.0029,
+      "step": 771
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 1.2334594178451425e-05,
+      "loss": 0.0062,
+      "step": 772
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 1.2231267729761487e-05,
+      "loss": 0.0047,
+      "step": 773
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 1.2128315538743646e-05,
+      "loss": 0.0059,
+      "step": 774
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 1.2025738625571026e-05,
+      "loss": 0.0057,
+      "step": 775
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 1.1923538006698154e-05,
+      "loss": 0.0047,
+      "step": 776
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 1.1821714694850689e-05,
+      "loss": 0.0035,
+      "step": 777
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 1.172026969901553e-05,
+      "loss": 0.0022,
+      "step": 778
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 1.161920402443077e-05,
+      "loss": 0.0052,
+      "step": 779
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 1.1518518672575701e-05,
+      "loss": 0.0025,
+      "step": 780
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 1.1418214641160958e-05,
+      "loss": 0.005,
+      "step": 781
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 1.1318292924118584e-05,
+      "loss": 0.0022,
+      "step": 782
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 1.121875451159221e-05,
+      "loss": 0.0023,
+      "step": 783
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 1.1119600389927182e-05,
+      "loss": 0.0029,
+      "step": 784
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 1.1020831541660915e-05,
+      "loss": 0.0056,
+      "step": 785
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 1.092244894551298e-05,
+      "loss": 0.0039,
+      "step": 786
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 1.0824453576375576e-05,
+      "loss": 0.0071,
+      "step": 787
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 1.0726846405303754e-05,
+      "loss": 0.0027,
+      "step": 788
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 1.062962839950587e-05,
+      "loss": 0.002,
+      "step": 789
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 1.0532800522333902e-05,
+      "loss": 0.0023,
+      "step": 790
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.0436363733274057e-05,
+      "loss": 0.0056,
+      "step": 791
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.0340318987937097e-05,
+      "loss": 0.0102,
+      "step": 792
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.0244667238048988e-05,
+      "loss": 0.0019,
+      "step": 793
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 1.014940943144142e-05,
+      "loss": 0.0044,
+      "step": 794
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 1.0054546512042424e-05,
+      "loss": 0.0017,
+      "step": 795
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 9.960079419866985e-06,
+      "loss": 0.0048,
+      "step": 796
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 9.866009091007833e-06,
+      "loss": 0.0021,
+      "step": 797
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 9.772336457626014e-06,
+      "loss": 0.012,
+      "step": 798
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 9.679062447941778e-06,
+      "loss": 0.006,
+      "step": 799
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 9.586187986225325e-06,
+      "loss": 0.0023,
+      "step": 800
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 9.493713992787672e-06,
+      "loss": 0.002,
+      "step": 801
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 9.401641383971477e-06,
+      "loss": 0.0019,
+      "step": 802
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 9.309971072142038e-06,
+      "loss": 0.0018,
+      "step": 803
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 9.218703965678204e-06,
+      "loss": 0.0049,
+      "step": 804
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 9.127840968963381e-06,
+      "loss": 0.0025,
+      "step": 805
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 9.03738298237658e-06,
+      "loss": 0.0041,
+      "step": 806
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 8.94733090228349e-06,
+      "loss": 0.0055,
+      "step": 807
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 8.857685621027568e-06,
+      "loss": 0.002,
+      "step": 808
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 8.768448026921245e-06,
+      "loss": 0.002,
+      "step": 809
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 8.67961900423711e-06,
+      "loss": 0.0038,
+      "step": 810
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 8.591199433199126e-06,
+      "loss": 0.0043,
+      "step": 811
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 8.503190189973914e-06,
+      "loss": 0.0017,
+      "step": 812
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 8.415592146662104e-06,
+      "loss": 0.0036,
+      "step": 813
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 8.328406171289621e-06,
+      "loss": 0.0044,
+      "step": 814
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 8.24163312779917e-06,
+      "loss": 0.0019,
+      "step": 815
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 8.155273876041614e-06,
+      "loss": 0.0029,
+      "step": 816
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 8.069329271767484e-06,
+      "loss": 0.0022,
+      "step": 817
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 7.983800166618482e-06,
+      "loss": 0.004,
+      "step": 818
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 7.898687408119065e-06,
+      "loss": 0.0028,
+      "step": 819
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 7.813991839667995e-06,
+      "loss": 0.0023,
+      "step": 820
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 7.72971430053005e-06,
+      "loss": 0.0019,
+      "step": 821
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 7.645855625827658e-06,
+      "loss": 0.0104,
+      "step": 822
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 7.56241664653266e-06,
+      "loss": 0.0031,
+      "step": 823
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 7.4793981894580034e-06,
+      "loss": 0.0024,
+      "step": 824
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 7.396801077249676e-06,
+      "loss": 0.0018,
+      "step": 825
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 7.3146261283784104e-06,
+      "loss": 0.0018,
+      "step": 826
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 7.2328741571316696e-06,
+      "loss": 0.0041,
+      "step": 827
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 7.1515459736055505e-06,
+      "loss": 0.0038,
+      "step": 828
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 7.070642383696763e-06,
+      "loss": 0.0031,
+      "step": 829
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 6.990164189094589e-06,
+      "loss": 0.0019,
+      "step": 830
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 6.910112187273066e-06,
+      "loss": 0.002,
+      "step": 831
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 6.830487171482935e-06,
+      "loss": 0.0083,
+      "step": 832
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 6.751289930743882e-06,
+      "loss": 0.0021,
+      "step": 833
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 6.6725212498366885e-06,
+      "loss": 0.0026,
+      "step": 834
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 6.594181909295427e-06,
+      "loss": 0.0078,
+      "step": 835
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 6.516272685399793e-06,
+      "loss": 0.0028,
+      "step": 836
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 6.438794350167337e-06,
+      "loss": 0.0019,
+      "step": 837
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 6.36174767134588e-06,
+      "loss": 0.004,
+      "step": 838
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 6.285133412405858e-06,
+      "loss": 0.0029,
+      "step": 839
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 6.208952332532786e-06,
+      "loss": 0.0036,
+      "step": 840
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 6.133205186619695e-06,
+      "loss": 0.002,
+      "step": 841
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 6.057892725259717e-06,
+      "loss": 0.0018,
+      "step": 842
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 5.983015694738597e-06,
+      "loss": 0.0048,
+      "step": 843
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 5.908574837027309e-06,
+      "loss": 0.0018,
+      "step": 844
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 5.83457088977471e-06,
+      "loss": 0.0029,
+      "step": 845
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 5.761004586300234e-06,
+      "loss": 0.0033,
+      "step": 846
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 5.687876655586583e-06,
+      "loss": 0.0026,
+      "step": 847
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 5.615187822272583e-06,
+      "loss": 0.0053,
+      "step": 848
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 5.542938806645931e-06,
+      "loss": 0.0022,
+      "step": 849
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 5.4711303246361144e-06,
+      "loss": 0.0058,
+      "step": 850
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 5.399763087807236e-06,
+      "loss": 0.0086,
+      "step": 851
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 5.328837803351083e-06,
+      "loss": 0.0035,
+      "step": 852
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 5.258355174079993e-06,
+      "loss": 0.0023,
+      "step": 853
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 5.188315898419971e-06,
+      "loss": 0.0023,
+      "step": 854
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 5.118720670403748e-06,
+      "loss": 0.0041,
+      "step": 855
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 5.04957017966391e-06,
+      "loss": 0.0045,
+      "step": 856
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 4.980865111426003e-06,
+      "loss": 0.0022,
+      "step": 857
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 4.912606146501886e-06,
+      "loss": 0.0025,
+      "step": 858
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 4.844793961282812e-06,
+      "loss": 0.0021,
+      "step": 859
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 4.777429227732844e-06,
+      "loss": 0.0024,
+      "step": 860
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 4.710512613382151e-06,
+      "loss": 0.0053,
+      "step": 861
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 4.644044781320422e-06,
+      "loss": 0.0021,
+      "step": 862
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 4.578026390190232e-06,
+      "loss": 0.002,
+      "step": 863
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 4.5124580941806165e-06,
+      "loss": 0.0039,
+      "step": 864
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 4.447340543020473e-06,
+      "loss": 0.0077,
+      "step": 865
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 4.382674381972224e-06,
+      "loss": 0.0018,
+      "step": 866
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 4.318460251825357e-06,
+      "loss": 0.002,
+      "step": 867
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 4.254698788890127e-06,
+      "loss": 0.0042,
+      "step": 868
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 4.191390624991159e-06,
+      "loss": 0.0024,
+      "step": 869
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 4.12853638746134e-06,
+      "loss": 0.0021,
+      "step": 870
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 4.0661366991354365e-06,
+      "loss": 0.0059,
+      "step": 871
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 4.004192178344029e-06,
+      "loss": 0.0064,
+      "step": 872
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 3.942703438907358e-06,
+      "loss": 0.0015,
+      "step": 873
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 3.881671090129247e-06,
+      "loss": 0.0044,
+      "step": 874
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 3.821095736791008e-06,
+      "loss": 0.0017,
+      "step": 875
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 3.7609779791455744e-06,
+      "loss": 0.0049,
+      "step": 876
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 3.7013184129113976e-06,
+      "loss": 0.0021,
+      "step": 877
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 3.6421176292666783e-06,
+      "loss": 0.002,
+      "step": 878
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 3.58337621484342e-06,
+      "loss": 0.0015,
+      "step": 879
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 3.5250947517216637e-06,
+      "loss": 0.0021,
+      "step": 880
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 3.4672738174236884e-06,
+      "loss": 0.0018,
+      "step": 881
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 3.4099139849083307e-06,
+      "loss": 0.0049,
+      "step": 882
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 3.353015822565253e-06,
+      "loss": 0.0022,
+      "step": 883
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 3.296579894209345e-06,
+      "loss": 0.0016,
+      "step": 884
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 3.2406067590751433e-06,
+      "loss": 0.0052,
+      "step": 885
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 3.1850969718112745e-06,
+      "loss": 0.0019,
+      "step": 886
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 3.1300510824749273e-06,
+      "loss": 0.0019,
+      "step": 887
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 3.0754696365265068e-06,
+      "loss": 0.0034,
+      "step": 888
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 3.0213531748240764e-06,
+      "loss": 0.0065,
+      "step": 889
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 2.9677022336181413e-06,
+      "loss": 0.0028,
+      "step": 890
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 2.914517344546258e-06,
+      "loss": 0.0024,
+      "step": 891
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 2.8617990346277657e-06,
+      "loss": 0.0019,
+      "step": 892
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 2.8095478262585907e-06,
+      "loss": 0.0038,
+      "step": 893
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 2.7577642372060673e-06,
+      "loss": 0.0024,
+      "step": 894
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 2.7064487806037985e-06,
+      "loss": 0.0085,
+      "step": 895
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 2.6556019649465525e-06,
+      "loss": 0.0016,
+      "step": 896
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 2.6052242940852787e-06,
+      "loss": 0.0019,
+      "step": 897
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 2.5553162672220465e-06,
+      "loss": 0.0021,
+      "step": 898
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 2.5058783789051467e-06,
+      "loss": 0.0041,
+      "step": 899
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 2.45691111902418e-06,
+      "loss": 0.0071,
+      "step": 900
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 2.4084149728051952e-06,
+      "loss": 0.0039,
+      "step": 901
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 2.360390420805869e-06,
+      "loss": 0.002,
+      "step": 902
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 2.3128379389108e-06,
+      "loss": 0.0015,
+      "step": 903
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 2.2657579983267064e-06,
+      "loss": 0.0016,
+      "step": 904
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 2.219151065577829e-06,
+      "loss": 0.0036,
+      "step": 905
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 2.1730176025012816e-06,
+      "loss": 0.0032,
+      "step": 906
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 2.1273580662424796e-06,
+      "loss": 0.0035,
+      "step": 907
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 2.082172909250568e-06,
+      "loss": 0.0022,
+      "step": 908
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 2.0374625792740464e-06,
+      "loss": 0.0067,
+      "step": 909
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 1.993227519356189e-06,
+      "loss": 0.0021,
+      "step": 910
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 1.9494681678307703e-06,
+      "loss": 0.004,
+      "step": 911
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 1.906184958317664e-06,
+      "loss": 0.0074,
+      "step": 912
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 1.8633783197185783e-06,
+      "loss": 0.004,
+      "step": 913
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 1.8210486762127499e-06,
+      "loss": 0.0031,
+      "step": 914
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 1.7791964472528232e-06,
+      "loss": 0.0037,
+      "step": 915
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 1.737822047560611e-06,
+      "loss": 0.0026,
+      "step": 916
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 1.696925887123052e-06,
+      "loss": 0.0033,
+      "step": 917
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 1.656508371188109e-06,
+      "loss": 0.0029,
+      "step": 918
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 1.6165699002607671e-06,
+      "loss": 0.0038,
+      "step": 919
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 1.5771108700990412e-06,
+      "loss": 0.0038,
+      "step": 920
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 1.538131671710108e-06,
+      "loss": 0.0018,
+      "step": 921
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 1.4996326913463754e-06,
+      "loss": 0.0041,
+      "step": 922
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 1.461614310501691e-06,
+      "loss": 0.0057,
+      "step": 923
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 1.4240769059075342e-06,
+      "loss": 0.004,
+      "step": 924
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 1.387020849529319e-06,
+      "loss": 0.0037,
+      "step": 925
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 1.3504465085626638e-06,
+      "loss": 0.0023,
+      "step": 926
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 1.3143542454297885e-06,
+      "loss": 0.0033,
+      "step": 927
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 1.2787444177759068e-06,
+      "loss": 0.0015,
+      "step": 928
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 1.243617378465689e-06,
+      "loss": 0.0014,
+      "step": 929
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 1.208973475579761e-06,
+      "loss": 0.004,
+      "step": 930
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 1.1748130524112666e-06,
+      "loss": 0.0039,
+      "step": 931
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 1.1411364474624264e-06,
+      "loss": 0.0033,
+      "step": 932
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 1.1079439944412406e-06,
+      "loss": 0.0035,
+      "step": 933
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 1.075236022258147e-06,
+      "loss": 0.004,
+      "step": 934
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 1.0430128550227625e-06,
+      "loss": 0.0019,
+      "step": 935
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 1.0112748120406856e-06,
+      "loss": 0.0022,
+      "step": 936
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 9.800222078103271e-07,
+      "loss": 0.0026,
+      "step": 937
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 9.492553520197733e-07,
+      "loss": 0.0016,
+      "step": 938
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 9.189745495437608e-07,
+      "loss": 0.0016,
+      "step": 939
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 8.891801004406119e-07,
+      "loss": 0.0025,
+      "step": 940
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 8.59872299949288e-07,
+      "loss": 0.0023,
+      "step": 941
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 8.31051438486441e-07,
+      "loss": 0.0023,
+      "step": 942
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 8.027178016435765e-07,
+      "loss": 0.0019,
+      "step": 943
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 7.748716701841685e-07,
+      "loss": 0.0024,
+      "step": 944
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 7.475133200409212e-07,
+      "loss": 0.002,
+      "step": 945
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 7.206430223130278e-07,
+      "loss": 0.0024,
+      "step": 946
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 6.9426104326345e-07,
+      "loss": 0.002,
+      "step": 947
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 6.683676443163311e-07,
+      "loss": 0.0016,
+      "step": 948
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 6.429630820543598e-07,
+      "loss": 0.0044,
+      "step": 949
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 6.180476082162656e-07,
+      "loss": 0.003,
+      "step": 950
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 5.936214696942887e-07,
+      "loss": 0.0024,
+      "step": 951
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 5.696849085317646e-07,
+      "loss": 0.0032,
+      "step": 952
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 5.462381619207091e-07,
+      "loss": 0.0047,
+      "step": 953
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 5.232814621994598e-07,
+      "loss": 0.0081,
+      "step": 954
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 5.008150368503994e-07,
+      "loss": 0.0021,
+      "step": 955
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 4.788391084976862e-07,
+      "loss": 0.0016,
+      "step": 956
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 4.573538949050327e-07,
+      "loss": 0.0017,
+      "step": 957
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 4.363596089735911e-07,
+      "loss": 0.0029,
+      "step": 958
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 4.1585645873978284e-07,
+      "loss": 0.0018,
+      "step": 959
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 3.958446473733002e-07,
+      "loss": 0.0029,
+      "step": 960
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 3.7632437317505207e-07,
+      "loss": 0.0015,
+      "step": 961
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 3.572958295752049e-07,
+      "loss": 0.0016,
+      "step": 962
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 3.387592051312782e-07,
+      "loss": 0.0053,
+      "step": 963
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 3.207146835262742e-07,
+      "loss": 0.0072,
+      "step": 964
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 3.0316244356683454e-07,
+      "loss": 0.0022,
+      "step": 965
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 2.8610265918151414e-07,
+      "loss": 0.0015,
+      "step": 966
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 2.695354994190047e-07,
+      "loss": 0.0021,
+      "step": 967
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 2.534611284465083e-07,
+      "loss": 0.0016,
+      "step": 968
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 2.3787970554806084e-07,
+      "loss": 0.0044,
+      "step": 969
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 2.2279138512300567e-07,
+      "loss": 0.002,
+      "step": 970
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 2.0819631668442253e-07,
+      "loss": 0.0025,
+      "step": 971
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 1.940946448576675e-07,
+      "loss": 0.0037,
+      "step": 972
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 1.8048650937893542e-07,
+      "loss": 0.0024,
+      "step": 973
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 1.6737204509387206e-07,
+      "loss": 0.0025,
+      "step": 974
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 1.5475138195623629e-07,
+      "loss": 0.0046,
+      "step": 975
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 1.4262464502663443e-07,
+      "loss": 0.0018,
+      "step": 976
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 1.309919544712268e-07,
+      "loss": 0.0066,
+      "step": 977
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 1.1985342556060652e-07,
+      "loss": 0.0026,
+      "step": 978
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 1.0920916866861142e-07,
+      "loss": 0.0054,
+      "step": 979
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 9.905928927123609e-08,
+      "loss": 0.0032,
+      "step": 980
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 8.940388794559939e-08,
+      "loss": 0.0031,
+      "step": 981
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 8.02430603689397e-08,
+      "loss": 0.0019,
+      "step": 982
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 7.157689731767669e-08,
+      "loss": 0.0029,
+      "step": 983
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 6.340548466648443e-08,
+      "loss": 0.0044,
+      "step": 984
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 5.572890338748082e-08,
+      "loss": 0.0057,
+      "step": 985
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 4.8547229549383844e-08,
+      "loss": 0.0018,
+      "step": 986
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 4.186053431680104e-08,
+      "loss": 0.0033,
+      "step": 987
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 3.566888394948009e-08,
+      "loss": 0.0032,
+      "step": 988
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 2.997233980168157e-08,
+      "loss": 0.0032,
+      "step": 989
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 2.4770958321568283e-08,
+      "loss": 0.0017,
+      "step": 990
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 2.0064791050633526e-08,
+      "loss": 0.0018,
+      "step": 991
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 1.5853884623195925e-08,
+      "loss": 0.0018,
+      "step": 992
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 1.2138280765944254e-08,
+      "loss": 0.0015,
+      "step": 993
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 8.918016297515541e-09,
+      "loss": 0.0031,
+      "step": 994
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 6.193123128134248e-09,
+      "loss": 0.0049,
+      "step": 995
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 3.963628259290308e-09,
+      "loss": 0.006,
+      "step": 996
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 2.229553783478222e-09,
+      "loss": 0.0022,
+      "step": 997
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 9.90916883986115e-10,
+      "loss": 0.0028,
+      "step": 998
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 2.477298346958978e-10,
+      "loss": 0.0022,
+      "step": 999
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 0.0,
+      "loss": 0.0029,
+      "step": 1000
+    }
+  ],
+  "max_steps": 1000,
+  "num_train_epochs": 13,
+  "total_flos": 7.311457550558822e+16,
+  "trial_name": null,
+  "trial_params": null
+}