diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,7318 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.996300863131936,
+  "eval_steps": 500,
+  "global_step": 1215,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.405405405405406e-07,
+      "loss": 0.5929,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0810810810810812e-06,
+      "loss": 0.5742,
+      "step": 2
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6216216216216219e-06,
+      "loss": 0.6021,
+      "step": 3
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1621621621621623e-06,
+      "loss": 0.5702,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.702702702702703e-06,
+      "loss": 0.6439,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2432432432432437e-06,
+      "loss": 0.5674,
+      "step": 6
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7837837837837844e-06,
+      "loss": 0.587,
+      "step": 7
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.324324324324325e-06,
+      "loss": 0.5865,
+      "step": 8
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.864864864864866e-06,
+      "loss": 0.599,
+      "step": 9
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.405405405405406e-06,
+      "loss": 0.5581,
+      "step": 10
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.945945945945947e-06,
+      "loss": 0.5865,
+      "step": 11
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.486486486486487e-06,
+      "loss": 0.5902,
+      "step": 12
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.027027027027028e-06,
+      "loss": 0.5336,
+      "step": 13
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.567567567567569e-06,
+      "loss": 0.5385,
+      "step": 14
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.108108108108109e-06,
+      "loss": 0.634,
+      "step": 15
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.64864864864865e-06,
+      "loss": 0.5324,
+      "step": 16
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.189189189189191e-06,
+      "loss": 0.5089,
+      "step": 17
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.729729729729732e-06,
+      "loss": 0.5402,
+      "step": 18
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.027027027027027e-05,
+      "loss": 0.4794,
+      "step": 19
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0810810810810812e-05,
+      "loss": 0.5166,
+      "step": 20
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1351351351351352e-05,
+      "loss": 0.5312,
+      "step": 21
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1891891891891894e-05,
+      "loss": 0.5118,
+      "step": 22
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2432432432432433e-05,
+      "loss": 0.4931,
+      "step": 23
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2972972972972975e-05,
+      "loss": 0.5146,
+      "step": 24
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3513513513513515e-05,
+      "loss": 0.4838,
+      "step": 25
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4054054054054055e-05,
+      "loss": 0.5099,
+      "step": 26
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4594594594594596e-05,
+      "loss": 0.5175,
+      "step": 27
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5135135135135138e-05,
+      "loss": 0.5304,
+      "step": 28
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5675675675675676e-05,
+      "loss": 0.5008,
+      "step": 29
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6216216216216218e-05,
+      "loss": 0.5069,
+      "step": 30
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6756756756756757e-05,
+      "loss": 0.4777,
+      "step": 31
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.72972972972973e-05,
+      "loss": 0.5198,
+      "step": 32
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.783783783783784e-05,
+      "loss": 0.5045,
+      "step": 33
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8378378378378383e-05,
+      "loss": 0.5328,
+      "step": 34
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.891891891891892e-05,
+      "loss": 0.5485,
+      "step": 35
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9459459459459463e-05,
+      "loss": 0.4978,
+      "step": 36
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 0.4765,
+      "step": 37
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999964438594985e-05,
+      "loss": 0.5169,
+      "step": 38
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999857754632864e-05,
+      "loss": 0.5228,
+      "step": 39
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999679948872396e-05,
+      "loss": 0.4622,
+      "step": 40
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999431022578194e-05,
+      "loss": 0.4699,
+      "step": 41
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999110977520687e-05,
+      "loss": 0.4748,
+      "step": 42
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999871981597613e-05,
+      "loss": 0.4532,
+      "step": 43
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999825754072657e-05,
+      "loss": 0.4658,
+      "step": 44
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997724155059836e-05,
+      "loss": 0.4711,
+      "step": 45
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997119662769523e-05,
+      "loss": 0.4783,
+      "step": 46
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999644406815495e-05,
+      "loss": 0.4677,
+      "step": 47
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995697376021125e-05,
+      "loss": 0.4806,
+      "step": 48
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994879591678744e-05,
+      "loss": 0.4708,
+      "step": 49
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993990720944115e-05,
+      "loss": 0.4972,
+      "step": 50
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999303077013914e-05,
+      "loss": 0.4815,
+      "step": 51
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999199974609125e-05,
+      "loss": 0.4754,
+      "step": 52
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990897656133385e-05,
+      "loss": 0.4962,
+      "step": 53
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989724508103916e-05,
+      "loss": 0.4393,
+      "step": 54
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9988480310346606e-05,
+      "loss": 0.4316,
+      "step": 55
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998716507171053e-05,
+      "loss": 0.4837,
+      "step": 56
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9985778801550044e-05,
+      "loss": 0.5239,
+      "step": 57
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9984321509724684e-05,
+      "loss": 0.4482,
+      "step": 58
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998279320659912e-05,
+      "loss": 0.4686,
+      "step": 59
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9981193903043077e-05,
+      "loss": 0.4664,
+      "step": 60
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997952361043125e-05,
+      "loss": 0.4734,
+      "step": 61
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997778234064323e-05,
+      "loss": 0.4188,
+      "step": 62
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9975970106063417e-05,
+      "loss": 0.506,
+      "step": 63
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9974086919580927e-05,
+      "loss": 0.4585,
+      "step": 64
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9972132794589515e-05,
+      "loss": 0.4934,
+      "step": 65
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9970107744987476e-05,
+      "loss": 0.4587,
+      "step": 66
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9968011785177515e-05,
+      "loss": 0.4429,
+      "step": 67
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.99658449300667e-05,
+      "loss": 0.5101,
+      "step": 68
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996360719506631e-05,
+      "loss": 0.4786,
+      "step": 69
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996129859609174e-05,
+      "loss": 0.4471,
+      "step": 70
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9958919149562405e-05,
+      "loss": 0.4665,
+      "step": 71
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9956468872401588e-05,
+      "loss": 0.4678,
+      "step": 72
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9953947782036353e-05,
+      "loss": 0.4268,
+      "step": 73
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.99513558963974e-05,
+      "loss": 0.5003,
+      "step": 74
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.994869323391895e-05,
+      "loss": 0.4517,
+      "step": 75
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9945959813538615e-05,
+      "loss": 0.4745,
+      "step": 76
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994315565469723e-05,
+      "loss": 0.4528,
+      "step": 77
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994028077733878e-05,
+      "loss": 0.4411,
+      "step": 78
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9937335201910187e-05,
+      "loss": 0.463,
+      "step": 79
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9934318949361218e-05,
+      "loss": 0.439,
+      "step": 80
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9931232041144308e-05,
+      "loss": 0.4864,
+      "step": 81
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992807449921441e-05,
+      "loss": 0.4645,
+      "step": 82
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9924846346028858e-05,
+      "loss": 0.5019,
+      "step": 83
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.992154760454718e-05,
+      "loss": 0.4351,
+      "step": 84
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9918178298230955e-05,
+      "loss": 0.4328,
+      "step": 85
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9914738451043633e-05,
+      "loss": 0.4475,
+      "step": 86
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9911228087450375e-05,
+      "loss": 0.4904,
+      "step": 87
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9907647232417873e-05,
+      "loss": 0.4788,
+      "step": 88
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9903995911414174e-05,
+      "loss": 0.4823,
+      "step": 89
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9900274150408503e-05,
+      "loss": 0.4874,
+      "step": 90
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9896481975871063e-05,
+      "loss": 0.4405,
+      "step": 91
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.989261941477287e-05,
+      "loss": 0.4489,
+      "step": 92
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9888686494585543e-05,
+      "loss": 0.4656,
+      "step": 93
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9884683243281117e-05,
+      "loss": 0.4468,
+      "step": 94
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9880609689331837e-05,
+      "loss": 0.4416,
+      "step": 95
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987646586170996e-05,
+      "loss": 0.4933,
+      "step": 96
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9872251789887563e-05,
+      "loss": 0.4968,
+      "step": 97
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.98679675038363e-05,
+      "loss": 0.4559,
+      "step": 98
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9863613034027224e-05,
+      "loss": 0.4765,
+      "step": 99
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9859188411430542e-05,
+      "loss": 0.4673,
+      "step": 100
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985469366751542e-05,
+      "loss": 0.4981,
+      "step": 101
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985012883424973e-05,
+      "loss": 0.4424,
+      "step": 102
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9845493944099856e-05,
+      "loss": 0.4242,
+      "step": 103
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.984078903003044e-05,
+      "loss": 0.454,
+      "step": 104
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9836014125504146e-05,
+      "loss": 0.5016,
+      "step": 105
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9831169264481446e-05,
+      "loss": 0.4823,
+      "step": 106
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9826254481420345e-05,
+      "loss": 0.4476,
+      "step": 107
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9821269811276164e-05,
+      "loss": 0.4405,
+      "step": 108
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981621528950128e-05,
+      "loss": 0.4729,
+      "step": 109
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9811090952044867e-05,
+      "loss": 0.454,
+      "step": 110
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.980589683535266e-05,
+      "loss": 0.4809,
+      "step": 111
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9800632976366668e-05,
+      "loss": 0.4367,
+      "step": 112
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9795299412524948e-05,
+      "loss": 0.4467,
+      "step": 113
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978989618176129e-05,
+      "loss": 0.5064,
+      "step": 114
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9784423322505e-05,
+      "loss": 0.471,
+      "step": 115
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9778880873680587e-05,
+      "loss": 0.4686,
+      "step": 116
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9773268874707504e-05,
+      "loss": 0.4623,
+      "step": 117
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9767587365499865e-05,
+      "loss": 0.4518,
+      "step": 118
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976183638646616e-05,
+      "loss": 0.4511,
+      "step": 119
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.975601597850897e-05,
+      "loss": 0.4487,
+      "step": 120
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.975012618302467e-05,
+      "loss": 0.447,
+      "step": 121
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974416704190314e-05,
+      "loss": 0.4655,
+      "step": 122
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9738138597527467e-05,
+      "loss": 0.4403,
+      "step": 123
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9732040892773644e-05,
+      "loss": 0.4687,
+      "step": 124
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9725873971010256e-05,
+      "loss": 0.4384,
+      "step": 125
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9719637876098186e-05,
+      "loss": 0.4577,
+      "step": 126
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9713332652390293e-05,
+      "loss": 0.4346,
+      "step": 127
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.97069583447311e-05,
+      "loss": 0.4425,
+      "step": 128
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9700514998456475e-05,
+      "loss": 0.4916,
+      "step": 129
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9694002659393306e-05,
+      "loss": 0.4236,
+      "step": 130
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9687421373859175e-05,
+      "loss": 0.441,
+      "step": 131
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9680771188662044e-05,
+      "loss": 0.4366,
+      "step": 132
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.967405215109989e-05,
+      "loss": 0.4591,
+      "step": 133
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.96672643089604e-05,
+      "loss": 0.4575,
+      "step": 134
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9660407710520613e-05,
+      "loss": 0.4006,
+      "step": 135
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9653482404546583e-05,
+      "loss": 0.5021,
+      "step": 136
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9646488440293035e-05,
+      "loss": 0.4576,
+      "step": 137
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9639425867503008e-05,
+      "loss": 0.4895,
+      "step": 138
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.96322947364075e-05,
+      "loss": 0.4275,
+      "step": 139
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9625095097725128e-05,
+      "loss": 0.5199,
+      "step": 140
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9617827002661734e-05,
+      "loss": 0.4943,
+      "step": 141
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9610490502910058e-05,
+      "loss": 0.4604,
+      "step": 142
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.960308565064935e-05,
+      "loss": 0.4723,
+      "step": 143
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9595612498544997e-05,
+      "loss": 0.4374,
+      "step": 144
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.958807109974816e-05,
+      "loss": 0.448,
+      "step": 145
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9580461507895382e-05,
+      "loss": 0.4349,
+      "step": 146
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.957278377710822e-05,
+      "loss": 0.4915,
+      "step": 147
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9565037961992853e-05,
+      "loss": 0.4812,
+      "step": 148
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.95572241176397e-05,
+      "loss": 0.4828,
+      "step": 149
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9549342299623012e-05,
+      "loss": 0.4482,
+      "step": 150
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.954139256400049e-05,
+      "loss": 0.458,
+      "step": 151
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9533374967312894e-05,
+      "loss": 0.4371,
+      "step": 152
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9525289566583626e-05,
+      "loss": 0.4712,
+      "step": 153
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9517136419318323e-05,
+      "loss": 0.4568,
+      "step": 154
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.950891558350446e-05,
+      "loss": 0.4172,
+      "step": 155
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.950062711761093e-05,
+      "loss": 0.4657,
+      "step": 156
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.949227108058764e-05,
+      "loss": 0.4531,
+      "step": 157
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9483847531865062e-05,
+      "loss": 0.4528,
+      "step": 158
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.947535653135385e-05,
+      "loss": 0.4487,
+      "step": 159
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9466798139444384e-05,
+      "loss": 0.4473,
+      "step": 160
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9458172417006347e-05,
+      "loss": 0.4796,
+      "step": 161
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9449479425388308e-05,
+      "loss": 0.469,
+      "step": 162
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9440719226417264e-05,
+      "loss": 0.4509,
+      "step": 163
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9431891882398207e-05,
+      "loss": 0.4698,
+      "step": 164
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9422997456113695e-05,
+      "loss": 0.4695,
+      "step": 165
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.941403601082338e-05,
+      "loss": 0.4476,
+      "step": 166
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9405007610263586e-05,
+      "loss": 0.501,
+      "step": 167
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.939591231864683e-05,
+      "loss": 0.4701,
+      "step": 168
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9386750200661386e-05,
+      "loss": 0.4602,
+      "step": 169
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9377521321470806e-05,
+      "loss": 0.4864,
+      "step": 170
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.936822574671348e-05,
+      "loss": 0.4183,
+      "step": 171
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9358863542502137e-05,
+      "loss": 0.48,
+      "step": 172
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9349434775423414e-05,
+      "loss": 0.4266,
+      "step": 173
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9339939512537347e-05,
+      "loss": 0.4488,
+      "step": 174
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9330377821376916e-05,
+      "loss": 0.4481,
+      "step": 175
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9320749769947555e-05,
+      "loss": 0.4173,
+      "step": 176
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.931105542672667e-05,
+      "loss": 0.4411,
+      "step": 177
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9301294860663152e-05,
+      "loss": 0.4841,
+      "step": 178
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9291468141176898e-05,
+      "loss": 0.4764,
+      "step": 179
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.928157533815829e-05,
+      "loss": 0.4394,
+      "step": 180
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9271616521967727e-05,
+      "loss": 0.4606,
+      "step": 181
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9261591763435106e-05,
+      "loss": 0.4798,
+      "step": 182
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9251501133859327e-05,
+      "loss": 0.4559,
+      "step": 183
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9241344705007784e-05,
+      "loss": 0.4423,
+      "step": 184
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9231122549115854e-05,
+      "loss": 0.4577,
+      "step": 185
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.922083473888638e-05,
+      "loss": 0.396,
+      "step": 186
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.921048134748916e-05,
+      "loss": 0.4322,
+      "step": 187
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9200062448560424e-05,
+      "loss": 0.4467,
+      "step": 188
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.918957811620231e-05,
+      "loss": 0.4636,
+      "step": 189
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.917902842498233e-05,
+      "loss": 0.4927,
+      "step": 190
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9168413449932856e-05,
+      "loss": 0.4133,
+      "step": 191
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9157733266550577e-05,
+      "loss": 0.4578,
+      "step": 192
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9146987950795953e-05,
+      "loss": 0.4455,
+      "step": 193
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.913617757909269e-05,
+      "loss": 0.4767,
+      "step": 194
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9125302228327194e-05,
+      "loss": 0.4529,
+      "step": 195
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9114361975848008e-05,
+      "loss": 0.4354,
+      "step": 196
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.910335689946529e-05,
+      "loss": 0.448,
+      "step": 197
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9092287077450227e-05,
+      "loss": 0.4678,
+      "step": 198
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9081152588534512e-05,
+      "loss": 0.4508,
+      "step": 199
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.906995351190976e-05,
+      "loss": 0.4593,
+      "step": 200
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9058689927226938e-05,
+      "loss": 0.4303,
+      "step": 201
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9047361914595836e-05,
+      "loss": 0.4588,
+      "step": 202
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9035969554584465e-05,
+      "loss": 0.458,
+      "step": 203
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9024512928218484e-05,
+      "loss": 0.4504,
+      "step": 204
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.901299211698064e-05,
+      "loss": 0.4267,
+      "step": 205
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9001407202810184e-05,
+      "loss": 0.4961,
+      "step": 206
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8989758268102276e-05,
+      "loss": 0.4809,
+      "step": 207
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.897804539570742e-05,
+      "loss": 0.4363,
+      "step": 208
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8966268668930847e-05,
+      "loss": 0.4589,
+      "step": 209
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.895442817153195e-05,
+      "loss": 0.4192,
+      "step": 210
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.894252398772368e-05,
+      "loss": 0.4707,
+      "step": 211
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8930556202171932e-05,
+      "loss": 0.4297,
+      "step": 212
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.891852489999496e-05,
+      "loss": 0.4415,
+      "step": 213
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8906430166762764e-05,
+      "loss": 0.4367,
+      "step": 214
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8894272088496488e-05,
+      "loss": 0.4341,
+      "step": 215
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.88820507516678e-05,
+      "loss": 0.4608,
+      "step": 216
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8869766243198288e-05,
+      "loss": 0.4348,
+      "step": 217
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.885741865045882e-05,
+      "loss": 0.4599,
+      "step": 218
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8845008061268948e-05,
+      "loss": 0.4483,
+      "step": 219
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8832534563896267e-05,
+      "loss": 0.4484,
+      "step": 220
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.88199982470558e-05,
+      "loss": 0.4679,
+      "step": 221
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8807399199909353e-05,
+      "loss": 0.4319,
+      "step": 222
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.879473751206489e-05,
+      "loss": 0.4672,
+      "step": 223
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8782013273575897e-05,
+      "loss": 0.44,
+      "step": 224
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8769226574940727e-05,
+      "loss": 0.4625,
+      "step": 225
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8756377507101976e-05,
+      "loss": 0.4808,
+      "step": 226
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8743466161445823e-05,
+      "loss": 0.4334,
+      "step": 227
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.873049262980139e-05,
+      "loss": 0.4283,
+      "step": 228
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.871745700444008e-05,
+      "loss": 0.4481,
+      "step": 229
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8704359378074922e-05,
+      "loss": 0.454,
+      "step": 230
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8691199843859917e-05,
+      "loss": 0.4588,
+      "step": 231
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8677978495389367e-05,
+      "loss": 0.4507,
+      "step": 232
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.866469542669722e-05,
+      "loss": 0.4912,
+      "step": 233
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8651350732256387e-05,
+      "loss": 0.4589,
+      "step": 234
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8637944506978093e-05,
+      "loss": 0.4317,
+      "step": 235
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.862447684621118e-05,
+      "loss": 0.4623,
+      "step": 236
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.861094784574143e-05,
+      "loss": 0.4605,
+      "step": 237
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8597357601790897e-05,
+      "loss": 0.4343,
+      "step": 238
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8583706211017218e-05,
+      "loss": 0.4227,
+      "step": 239
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8569993770512917e-05,
+      "loss": 0.4464,
+      "step": 240
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8556220377804726e-05,
+      "loss": 0.4278,
+      "step": 241
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8542386130852884e-05,
+      "loss": 0.4712,
+      "step": 242
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8528491128050442e-05,
+      "loss": 0.3954,
+      "step": 243
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8514535468222567e-05,
+      "loss": 0.4589,
+      "step": 244
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8500519250625833e-05,
+      "loss": 0.4513,
+      "step": 245
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.848644257494751e-05,
+      "loss": 0.458,
+      "step": 246
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8472305541304882e-05,
+      "loss": 0.5164,
+      "step": 247
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.84581082502445e-05,
+      "loss": 0.4607,
+      "step": 248
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8443850802741487e-05,
+      "loss": 0.4729,
+      "step": 249
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8429533300198818e-05,
+      "loss": 0.4592,
+      "step": 250
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8415155844446593e-05,
+      "loss": 0.4316,
+      "step": 251
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8400718537741317e-05,
+      "loss": 0.447,
+      "step": 252
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.838622148276517e-05,
+      "loss": 0.4461,
+      "step": 253
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8371664782625287e-05,
+      "loss": 0.4425,
+      "step": 254
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8357048540853005e-05,
+      "loss": 0.4226,
+      "step": 255
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8342372861403145e-05,
+      "loss": 0.4227,
+      "step": 256
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8327637848653262e-05,
+      "loss": 0.4892,
+      "step": 257
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8312843607402908e-05,
+      "loss": 0.4414,
+      "step": 258
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8297990242872883e-05,
+      "loss": 0.4575,
+      "step": 259
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.828307786070449e-05,
+      "loss": 0.452,
+      "step": 260
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8268106566958784e-05,
+      "loss": 0.4269,
+      "step": 261
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.825307646811581e-05,
+      "loss": 0.4253,
+      "step": 262
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8237987671073848e-05,
+      "loss": 0.4895,
+      "step": 263
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8222840283148678e-05,
+      "loss": 0.4219,
+      "step": 264
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8207634412072765e-05,
+      "loss": 0.3996,
+      "step": 265
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8192370165994548e-05,
+      "loss": 0.4657,
+      "step": 266
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8177047653477622e-05,
+      "loss": 0.4486,
+      "step": 267
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8161666983500014e-05,
+      "loss": 0.4682,
+      "step": 268
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8146228265453365e-05,
+      "loss": 0.4328,
+      "step": 269
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.813073160914218e-05,
+      "loss": 0.4484,
+      "step": 270
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8115177124783026e-05,
+      "loss": 0.4604,
+      "step": 271
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.809956492300377e-05,
+      "loss": 0.4257,
+      "step": 272
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.808389511484278e-05,
+      "loss": 0.445,
+      "step": 273
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8068167811748134e-05,
+      "loss": 0.4006,
+      "step": 274
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8052383125576833e-05,
+      "loss": 0.4238,
+      "step": 275
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8036541168593998e-05,
+      "loss": 0.4159,
+      "step": 276
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.802064205347208e-05,
+      "loss": 0.4613,
+      "step": 277
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.800468589329005e-05,
+      "loss": 0.4738,
+      "step": 278
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7988672801532603e-05,
+      "loss": 0.4251,
+      "step": 279
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7972602892089354e-05,
+      "loss": 0.4568,
+      "step": 280
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.795647627925401e-05,
+      "loss": 0.4563,
+      "step": 281
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7940293077723573e-05,
+      "loss": 0.4776,
+      "step": 282
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7924053402597523e-05,
+      "loss": 0.4405,
+      "step": 283
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7907757369376984e-05,
+      "loss": 0.4139,
+      "step": 284
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.789140509396394e-05,
+      "loss": 0.4858,
+      "step": 285
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7874996692660352e-05,
+      "loss": 0.4085,
+      "step": 286
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7858532282167386e-05,
+      "loss": 0.435,
+      "step": 287
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.784201197958456e-05,
+      "loss": 0.4393,
+      "step": 288
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7825435902408905e-05,
+      "loss": 0.5023,
+      "step": 289
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.780880416853414e-05,
+      "loss": 0.4446,
+      "step": 290
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7792116896249833e-05,
+      "loss": 0.4706,
+      "step": 291
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.777537420424055e-05,
+      "loss": 0.4514,
+      "step": 292
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7758576211585023e-05,
+      "loss": 0.4736,
+      "step": 293
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7741723037755294e-05,
+      "loss": 0.4394,
+      "step": 294
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.772481480261587e-05,
+      "loss": 0.4562,
+      "step": 295
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7707851626422877e-05,
+      "loss": 0.4572,
+      "step": 296
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7690833629823186e-05,
+      "loss": 0.4494,
+      "step": 297
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.767376093385357e-05,
+      "loss": 0.3927,
+      "step": 298
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7656633659939846e-05,
+      "loss": 0.4372,
+      "step": 299
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.763945192989599e-05,
+      "loss": 0.4555,
+      "step": 300
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7622215865923303e-05,
+      "loss": 0.4241,
+      "step": 301
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.760492559060951e-05,
+      "loss": 0.4456,
+      "step": 302
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.758758122692791e-05,
+      "loss": 0.4368,
+      "step": 303
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7570182898236493e-05,
+      "loss": 0.4235,
+      "step": 304
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7552730728277052e-05,
+      "loss": 0.4544,
+      "step": 305
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7535224841174336e-05,
+      "loss": 0.4045,
+      "step": 306
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.751766536143513e-05,
+      "loss": 0.4487,
+      "step": 307
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.750005241394738e-05,
+      "loss": 0.4811,
+      "step": 308
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7482386123979327e-05,
+      "loss": 0.4704,
+      "step": 309
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7464666617178586e-05,
+      "loss": 0.3911,
+      "step": 310
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7446894019571273e-05,
+      "loss": 0.4307,
+      "step": 311
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.742906845756109e-05,
+      "loss": 0.495,
+      "step": 312
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7411190057928444e-05,
+      "loss": 0.4512,
+      "step": 313
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.739325894782954e-05,
+      "loss": 0.427,
+      "step": 314
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7375275254795474e-05,
+      "loss": 0.4341,
+      "step": 315
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.735723910673132e-05,
+      "loss": 0.518,
+      "step": 316
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.733915063191523e-05,
+      "loss": 0.4755,
+      "step": 317
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.732100995899752e-05,
+      "loss": 0.4295,
+      "step": 318
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7302817216999758e-05,
+      "loss": 0.4233,
+      "step": 319
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7284572535313836e-05,
+      "loss": 0.4547,
+      "step": 320
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7266276043701054e-05,
+      "loss": 0.4245,
+      "step": 321
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.72479278722912e-05,
+      "loss": 0.4444,
+      "step": 322
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7229528151581633e-05,
+      "loss": 0.4731,
+      "step": 323
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.721107701243633e-05,
+      "loss": 0.4477,
+      "step": 324
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.719257458608498e-05,
+      "loss": 0.4633,
+      "step": 325
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7174021004122038e-05,
+      "loss": 0.4235,
+      "step": 326
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7155416398505795e-05,
+      "loss": 0.4063,
+      "step": 327
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.713676090155743e-05,
+      "loss": 0.5046,
+      "step": 328
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.711805464596008e-05,
+      "loss": 0.4608,
+      "step": 329
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7099297764757895e-05,
+      "loss": 0.4198,
+      "step": 330
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7080490391355083e-05,
+      "loss": 0.4434,
+      "step": 331
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7061632659514967e-05,
+      "loss": 0.4507,
+      "step": 332
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7042724703359033e-05,
+      "loss": 0.4151,
+      "step": 333
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7023766657365985e-05,
+      "loss": 0.445,
+      "step": 334
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.700475865637077e-05,
+      "loss": 0.4697,
+      "step": 335
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.698570083556363e-05,
+      "loss": 0.4465,
+      "step": 336
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6966593330489148e-05,
+      "loss": 0.5014,
+      "step": 337
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.694743627704526e-05,
+      "loss": 0.4307,
+      "step": 338
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.692822981148232e-05,
+      "loss": 0.4406,
+      "step": 339
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.690897407040211e-05,
+      "loss": 0.4564,
+      "step": 340
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.688966919075687e-05,
+      "loss": 0.4353,
+      "step": 341
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.687031530984832e-05,
+      "loss": 0.456,
+      "step": 342
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.685091256532671e-05,
+      "loss": 0.4706,
+      "step": 343
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.683146109518981e-05,
+      "loss": 0.4585,
+      "step": 344
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6811961037781936e-05,
+      "loss": 0.4371,
+      "step": 345
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6792412531792985e-05,
+      "loss": 0.4866,
+      "step": 346
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6772815716257414e-05,
+      "loss": 0.4347,
+      "step": 347
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6753170730553288e-05,
+      "loss": 0.4406,
+      "step": 348
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6733477714401264e-05,
+      "loss": 0.4457,
+      "step": 349
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6713736807863607e-05,
+      "loss": 0.437,
+      "step": 350
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6693948151343193e-05,
+      "loss": 0.4177,
+      "step": 351
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6674111885582505e-05,
+      "loss": 0.4561,
+      "step": 352
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6654228151662642e-05,
+      "loss": 0.4357,
+      "step": 353
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6634297091002305e-05,
+      "loss": 0.4473,
+      "step": 354
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.66143188453568e-05,
+      "loss": 0.4428,
+      "step": 355
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6594293556817022e-05,
+      "loss": 0.475,
+      "step": 356
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6574221367808454e-05,
+      "loss": 0.4178,
+      "step": 357
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6554102421090138e-05,
+      "loss": 0.4401,
+      "step": 358
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.653393685975368e-05,
+      "loss": 0.4531,
+      "step": 359
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6513724827222225e-05,
+      "loss": 0.4551,
+      "step": 360
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6493466467249416e-05,
+      "loss": 0.4631,
+      "step": 361
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.647316192391841e-05,
+      "loss": 0.4564,
+      "step": 362
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6452811341640825e-05,
+      "loss": 0.4327,
+      "step": 363
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6432414865155712e-05,
+      "loss": 0.4521,
+      "step": 364
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6411972639528555e-05,
+      "loss": 0.4281,
+      "step": 365
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.63914848101502e-05,
+      "loss": 0.4285,
+      "step": 366
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.637095152273585e-05,
+      "loss": 0.415,
+      "step": 367
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.635037292332401e-05,
+      "loss": 0.4731,
+      "step": 368
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6329749158275466e-05,
+      "loss": 0.4438,
+      "step": 369
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6309080374272233e-05,
+      "loss": 0.4335,
+      "step": 370
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6288366718316504e-05,
+      "loss": 0.4229,
+      "step": 371
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6267608337729622e-05,
+      "loss": 0.4286,
+      "step": 372
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6246805380151028e-05,
+      "loss": 0.4488,
+      "step": 373
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.62259579935372e-05,
+      "loss": 0.4079,
+      "step": 374
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.620506632616061e-05,
+      "loss": 0.4251,
+      "step": 375
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.618413052660866e-05,
+      "loss": 0.4148,
+      "step": 376
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6163150743782648e-05,
+      "loss": 0.4976,
+      "step": 377
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6142127126896682e-05,
+      "loss": 0.3998,
+      "step": 378
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.612105982547663e-05,
+      "loss": 0.4401,
+      "step": 379
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6099948989359063e-05,
+      "loss": 0.4221,
+      "step": 380
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6078794768690182e-05,
+      "loss": 0.4081,
+      "step": 381
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6057597313924747e-05,
+      "loss": 0.4433,
+      "step": 382
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.603635677582501e-05,
+      "loss": 0.4385,
+      "step": 383
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.601507330545965e-05,
+      "loss": 0.4457,
+      "step": 384
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5993747054202683e-05,
+      "loss": 0.4242,
+      "step": 385
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5972378173732407e-05,
+      "loss": 0.4766,
+      "step": 386
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5950966816030305e-05,
+      "loss": 0.4442,
+      "step": 387
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5929513133379967e-05,
+      "loss": 0.446,
+      "step": 388
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.590801727836601e-05,
+      "loss": 0.4965,
+      "step": 389
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5886479403873004e-05,
+      "loss": 0.4266,
+      "step": 390
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5864899663084356e-05,
+      "loss": 0.4362,
+      "step": 391
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5843278209481247e-05,
+      "loss": 0.4599,
+      "step": 392
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5821615196841535e-05,
+      "loss": 0.4164,
+      "step": 393
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5799910779238654e-05,
+      "loss": 0.444,
+      "step": 394
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5778165111040512e-05,
+      "loss": 0.4587,
+      "step": 395
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5756378346908422e-05,
+      "loss": 0.4207,
+      "step": 396
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.573455064179597e-05,
+      "loss": 0.4269,
+      "step": 397
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5712682150947926e-05,
+      "loss": 0.4276,
+      "step": 398
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5690773029899143e-05,
+      "loss": 0.4001,
+      "step": 399
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5668823434473446e-05,
+      "loss": 0.4569,
+      "step": 400
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5646833520782525e-05,
+      "loss": 0.4334,
+      "step": 401
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.562480344522483e-05,
+      "loss": 0.4708,
+      "step": 402
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5602733364484443e-05,
+      "loss": 0.4572,
+      "step": 403
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5580623435529983e-05,
+      "loss": 0.4658,
+      "step": 404
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5558473815613476e-05,
+      "loss": 0.4032,
+      "step": 405
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5536284662269246e-05,
+      "loss": 0.4203,
+      "step": 406
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.551405613331278e-05,
+      "loss": 0.425,
+      "step": 407
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5491788386839635e-05,
+      "loss": 0.4501,
+      "step": 408
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5469481581224274e-05,
+      "loss": 0.4331,
+      "step": 409
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.544713587511896e-05,
+      "loss": 0.4282,
+      "step": 410
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5424751427452644e-05,
+      "loss": 0.4306,
+      "step": 411
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5402328397429797e-05,
+      "loss": 0.413,
+      "step": 412
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.537986694452932e-05,
+      "loss": 0.4066,
+      "step": 413
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.535736722850338e-05,
+      "loss": 0.4295,
+      "step": 414
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5334829409376272e-05,
+      "loss": 0.4111,
+      "step": 415
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.53122536474433e-05,
+      "loss": 0.4473,
+      "step": 416
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5289640103269626e-05,
+      "loss": 0.4043,
+      "step": 417
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5266988937689134e-05,
+      "loss": 0.394,
+      "step": 418
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5244300311803278e-05,
+      "loss": 0.4433,
+      "step": 419
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5221574386979938e-05,
+      "loss": 0.4351,
+      "step": 420
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5198811324852278e-05,
+      "loss": 0.4308,
+      "step": 421
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5176011287317601e-05,
+      "loss": 0.3722,
+      "step": 422
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5153174436536166e-05,
+      "loss": 0.3858,
+      "step": 423
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5130300934930083e-05,
+      "loss": 0.3864,
+      "step": 424
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5107390945182119e-05,
+      "loss": 0.3844,
+      "step": 425
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5084444630234562e-05,
+      "loss": 0.3906,
+      "step": 426
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.506146215328805e-05,
+      "loss": 0.3693,
+      "step": 427
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5038443677800415e-05,
+      "loss": 0.393,
+      "step": 428
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.501538936748553e-05,
+      "loss": 0.3629,
+      "step": 429
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4992299386312121e-05,
+      "loss": 0.3512,
+      "step": 430
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4969173898502625e-05,
+      "loss": 0.3685,
+      "step": 431
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.494601306853201e-05,
+      "loss": 0.3948,
+      "step": 432
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4922817061126607e-05,
+      "loss": 0.4002,
+      "step": 433
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4899586041262936e-05,
+      "loss": 0.3701,
+      "step": 434
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4876320174166544e-05,
+      "loss": 0.3645,
+      "step": 435
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4853019625310813e-05,
+      "loss": 0.3703,
+      "step": 436
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.482968456041579e-05,
+      "loss": 0.345,
+      "step": 437
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.480631514544702e-05,
+      "loss": 0.389,
+      "step": 438
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4782911546614344e-05,
+      "loss": 0.3613,
+      "step": 439
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4759473930370738e-05,
+      "loss": 0.3994,
+      "step": 440
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.473600246341111e-05,
+      "loss": 0.371,
+      "step": 441
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.471249731267113e-05,
+      "loss": 0.3577,
+      "step": 442
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4688958645326042e-05,
+      "loss": 0.347,
+      "step": 443
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.466538662878945e-05,
+      "loss": 0.375,
+      "step": 444
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.464178143071217e-05,
+      "loss": 0.353,
+      "step": 445
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4618143218980998e-05,
+      "loss": 0.3518,
+      "step": 446
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4594472161717538e-05,
+      "loss": 0.3366,
+      "step": 447
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4570768427277009e-05,
+      "loss": 0.3521,
+      "step": 448
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4547032184247022e-05,
+      "loss": 0.3273,
+      "step": 449
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4523263601446411e-05,
+      "loss": 0.3558,
+      "step": 450
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4499462847924016e-05,
+      "loss": 0.3236,
+      "step": 451
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4475630092957483e-05,
+      "loss": 0.372,
+      "step": 452
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4451765506052064e-05,
+      "loss": 0.3545,
+      "step": 453
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4427869256939403e-05,
+      "loss": 0.3291,
+      "step": 454
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4403941515576344e-05,
+      "loss": 0.3941,
+      "step": 455
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4379982452143706e-05,
+      "loss": 0.3824,
+      "step": 456
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4355992237045078e-05,
+      "loss": 0.3494,
+      "step": 457
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4331971040905615e-05,
+      "loss": 0.3648,
+      "step": 458
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.430791903457081e-05,
+      "loss": 0.3413,
+      "step": 459
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4283836389105293e-05,
+      "loss": 0.331,
+      "step": 460
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4259723275791606e-05,
+      "loss": 0.3274,
+      "step": 461
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4235579866128985e-05,
+      "loss": 0.3613,
+      "step": 462
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4211406331832147e-05,
+      "loss": 0.3535,
+      "step": 463
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.418720284483006e-05,
+      "loss": 0.3456,
+      "step": 464
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.416296957726472e-05,
+      "loss": 0.3476,
+      "step": 465
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4138706701489945e-05,
+      "loss": 0.3464,
+      "step": 466
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4114414390070112e-05,
+      "loss": 0.3131,
+      "step": 467
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4090092815778977e-05,
+      "loss": 0.3591,
+      "step": 468
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.406574215159841e-05,
+      "loss": 0.3598,
+      "step": 469
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4041362570717171e-05,
+      "loss": 0.3468,
+      "step": 470
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4016954246529697e-05,
+      "loss": 0.3672,
+      "step": 471
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3992517352634844e-05,
+      "loss": 0.3126,
+      "step": 472
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3968052062834667e-05,
+      "loss": 0.3343,
+      "step": 473
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3943558551133186e-05,
+      "loss": 0.358,
+      "step": 474
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.391903699173514e-05,
+      "loss": 0.3516,
+      "step": 475
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3894487559044745e-05,
+      "loss": 0.3239,
+      "step": 476
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3869910427664467e-05,
+      "loss": 0.3336,
+      "step": 477
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.384530577239377e-05,
+      "loss": 0.3423,
+      "step": 478
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.382067376822788e-05,
+      "loss": 0.3448,
+      "step": 479
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3796014590356524e-05,
+      "loss": 0.363,
+      "step": 480
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3771328414162714e-05,
+      "loss": 0.3417,
+      "step": 481
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3746615415221466e-05,
+      "loss": 0.3342,
+      "step": 482
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3721875769298575e-05,
+      "loss": 0.3124,
+      "step": 483
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3697109652349354e-05,
+      "loss": 0.3413,
+      "step": 484
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3672317240517388e-05,
+      "loss": 0.3285,
+      "step": 485
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3647498710133274e-05,
+      "loss": 0.3723,
+      "step": 486
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3622654237713371e-05,
+      "loss": 0.3416,
+      "step": 487
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3597783999958556e-05,
+      "loss": 0.3558,
+      "step": 488
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3572888173752947e-05,
+      "loss": 0.3468,
+      "step": 489
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3547966936162648e-05,
+      "loss": 0.3169,
+      "step": 490
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3523020464434515e-05,
+      "loss": 0.3202,
+      "step": 491
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.349804893599486e-05,
+      "loss": 0.3505,
+      "step": 492
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3473052528448203e-05,
+      "loss": 0.3709,
+      "step": 493
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3448031419576031e-05,
+      "loss": 0.3479,
+      "step": 494
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3422985787335494e-05,
+      "loss": 0.3826,
+      "step": 495
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3397915809858169e-05,
+      "loss": 0.3414,
+      "step": 496
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3372821665448776e-05,
+      "loss": 0.3402,
+      "step": 497
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3347703532583928e-05,
+      "loss": 0.3257,
+      "step": 498
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3322561589910841e-05,
+      "loss": 0.3357,
+      "step": 499
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3297396016246077e-05,
+      "loss": 0.3585,
+      "step": 500
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3272206990574264e-05,
+      "loss": 0.3639,
+      "step": 501
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3246994692046837e-05,
+      "loss": 0.3767,
+      "step": 502
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.322175929998074e-05,
+      "loss": 0.3586,
+      "step": 503
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3196500993857182e-05,
+      "loss": 0.3624,
+      "step": 504
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3171219953320331e-05,
+      "loss": 0.354,
+      "step": 505
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3145916358176045e-05,
+      "loss": 0.372,
+      "step": 506
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.312059038839061e-05,
+      "loss": 0.3337,
+      "step": 507
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3095242224089435e-05,
+      "loss": 0.3226,
+      "step": 508
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3069872045555792e-05,
+      "loss": 0.3439,
+      "step": 509
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3044480033229515e-05,
+      "loss": 0.3529,
+      "step": 510
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3019066367705734e-05,
+      "loss": 0.37,
+      "step": 511
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2993631229733584e-05,
+      "loss": 0.3616,
+      "step": 512
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2968174800214901e-05,
+      "loss": 0.3414,
+      "step": 513
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2942697260202976e-05,
+      "loss": 0.3333,
+      "step": 514
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2917198790901231e-05,
+      "loss": 0.3417,
+      "step": 515
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2891679573661939e-05,
+      "loss": 0.3371,
+      "step": 516
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2866139789984953e-05,
+      "loss": 0.3441,
+      "step": 517
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2840579621516383e-05,
+      "loss": 0.3323,
+      "step": 518
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2814999250047335e-05,
+      "loss": 0.3736,
+      "step": 519
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2789398857512598e-05,
+      "loss": 0.3703,
+      "step": 520
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2763778625989355e-05,
+      "loss": 0.3706,
+      "step": 521
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2738138737695895e-05,
+      "loss": 0.3619,
+      "step": 522
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2712479374990303e-05,
+      "loss": 0.3479,
+      "step": 523
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2686800720369186e-05,
+      "loss": 0.3621,
+      "step": 524
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2661102956466345e-05,
+      "loss": 0.3171,
+      "step": 525
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.26353862660515e-05,
+      "loss": 0.35,
+      "step": 526
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2609650832028979e-05,
+      "loss": 0.3521,
+      "step": 527
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.258389683743642e-05,
+      "loss": 0.3289,
+      "step": 528
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.255812446544347e-05,
+      "loss": 0.3407,
+      "step": 529
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2532333899350482e-05,
+      "loss": 0.3671,
+      "step": 530
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2506525322587207e-05,
+      "loss": 0.3517,
+      "step": 531
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2480698918711496e-05,
+      "loss": 0.3434,
+      "step": 532
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2454854871407993e-05,
+      "loss": 0.3205,
+      "step": 533
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2428993364486824e-05,
+      "loss": 0.3653,
+      "step": 534
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.240311458188229e-05,
+      "loss": 0.3441,
+      "step": 535
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2377218707651563e-05,
+      "loss": 0.3502,
+      "step": 536
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2351305925973387e-05,
+      "loss": 0.3236,
+      "step": 537
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.232537642114674e-05,
+      "loss": 0.3334,
+      "step": 538
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2299430377589548e-05,
+      "loss": 0.3465,
+      "step": 539
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2273467979837363e-05,
+      "loss": 0.3445,
+      "step": 540
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2247489412542055e-05,
+      "loss": 0.3785,
+      "step": 541
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2221494860470493e-05,
+      "loss": 0.3203,
+      "step": 542
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2195484508503235e-05,
+      "loss": 0.3791,
+      "step": 543
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2169458541633217e-05,
+      "loss": 0.3425,
+      "step": 544
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2143417144964425e-05,
+      "loss": 0.3616,
+      "step": 545
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.211736050371059e-05,
+      "loss": 0.3878,
+      "step": 546
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.209128880319387e-05,
+      "loss": 0.3514,
+      "step": 547
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2065202228843525e-05,
+      "loss": 0.3444,
+      "step": 548
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2039100966194594e-05,
+      "loss": 0.3674,
+      "step": 549
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2012985200886602e-05,
+      "loss": 0.3399,
+      "step": 550
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1986855118662207e-05,
+      "loss": 0.3408,
+      "step": 551
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1960710905365893e-05,
+      "loss": 0.3551,
+      "step": 552
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1934552746942653e-05,
+      "loss": 0.386,
+      "step": 553
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1908380829436668e-05,
+      "loss": 0.3776,
+      "step": 554
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1882195338989959e-05,
+      "loss": 0.3426,
+      "step": 555
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1855996461841095e-05,
+      "loss": 0.378,
+      "step": 556
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1829784384323857e-05,
+      "loss": 0.336,
+      "step": 557
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.18035592928659e-05,
+      "loss": 0.3517,
+      "step": 558
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1777321373987448e-05,
+      "loss": 0.3666,
+      "step": 559
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.175107081429995e-05,
+      "loss": 0.3396,
+      "step": 560
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1724807800504768e-05,
+      "loss": 0.3572,
+      "step": 561
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1698532519391827e-05,
+      "loss": 0.3459,
+      "step": 562
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1672245157838318e-05,
+      "loss": 0.3473,
+      "step": 563
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.164594590280734e-05,
+      "loss": 0.3386,
+      "step": 564
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1619634941346587e-05,
+      "loss": 0.3506,
+      "step": 565
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1593312460587005e-05,
+      "loss": 0.3555,
+      "step": 566
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.156697864774148e-05,
+      "loss": 0.361,
+      "step": 567
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1540633690103488e-05,
+      "loss": 0.3427,
+      "step": 568
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1514277775045768e-05,
+      "loss": 0.3509,
+      "step": 569
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1487911090018995e-05,
+      "loss": 0.3519,
+      "step": 570
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1461533822550443e-05,
+      "loss": 0.3484,
+      "step": 571
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1435146160242648e-05,
+      "loss": 0.3648,
+      "step": 572
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.140874829077208e-05,
+      "loss": 0.3737,
+      "step": 573
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1382340401887808e-05,
+      "loss": 0.3454,
+      "step": 574
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1355922681410155e-05,
+      "loss": 0.3634,
+      "step": 575
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1329495317229374e-05,
+      "loss": 0.3615,
+      "step": 576
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1303058497304304e-05,
+      "loss": 0.3319,
+      "step": 577
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1276612409661037e-05,
+      "loss": 0.3609,
+      "step": 578
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.125015724239158e-05,
+      "loss": 0.3206,
+      "step": 579
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1223693183652518e-05,
+      "loss": 0.3446,
+      "step": 580
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1197220421663675e-05,
+      "loss": 0.3405,
+      "step": 581
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1170739144706766e-05,
+      "loss": 0.3514,
+      "step": 582
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.114424954112408e-05,
+      "loss": 0.3545,
+      "step": 583
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.111775179931712e-05,
+      "loss": 0.3618,
+      "step": 584
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.109124610774527e-05,
+      "loss": 0.3603,
+      "step": 585
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1064732654924462e-05,
+      "loss": 0.3635,
+      "step": 586
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1038211629425816e-05,
+      "loss": 0.3899,
+      "step": 587
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1011683219874324e-05,
+      "loss": 0.3565,
+      "step": 588
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0985147614947484e-05,
+      "loss": 0.3543,
+      "step": 589
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0958605003373977e-05,
+      "loss": 0.359,
+      "step": 590
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0932055573932317e-05,
+      "loss": 0.3298,
+      "step": 591
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0905499515449501e-05,
+      "loss": 0.3351,
+      "step": 592
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0878937016799684e-05,
+      "loss": 0.3288,
+      "step": 593
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.085236826690282e-05,
+      "loss": 0.3401,
+      "step": 594
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0825793454723325e-05,
+      "loss": 0.389,
+      "step": 595
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0799212769268727e-05,
+      "loss": 0.3284,
+      "step": 596
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0772626399588337e-05,
+      "loss": 0.3474,
+      "step": 597
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0746034534771881e-05,
+      "loss": 0.363,
+      "step": 598
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.071943736394817e-05,
+      "loss": 0.3579,
+      "step": 599
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.069283507628377e-05,
+      "loss": 0.3586,
+      "step": 600
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0666227860981615e-05,
+      "loss": 0.3647,
+      "step": 601
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.06396159072797e-05,
+      "loss": 0.3368,
+      "step": 602
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0612999404449722e-05,
+      "loss": 0.3658,
+      "step": 603
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0586378541795723e-05,
+      "loss": 0.3579,
+      "step": 604
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.055975350865276e-05,
+      "loss": 0.3721,
+      "step": 605
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0533124494385552e-05,
+      "loss": 0.3509,
+      "step": 606
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0506491688387128e-05,
+      "loss": 0.3258,
+      "step": 607
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0479855280077493e-05,
+      "loss": 0.3703,
+      "step": 608
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0453215458902264e-05,
+      "loss": 0.366,
+      "step": 609
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0426572414331338e-05,
+      "loss": 0.3328,
+      "step": 610
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0399926335857532e-05,
+      "loss": 0.3713,
+      "step": 611
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0373277412995242e-05,
+      "loss": 0.3792,
+      "step": 612
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0346625835279103e-05,
+      "loss": 0.3608,
+      "step": 613
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.031997179226262e-05,
+      "loss": 0.3465,
+      "step": 614
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0293315473516833e-05,
+      "loss": 0.3339,
+      "step": 615
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0266657068628983e-05,
+      "loss": 0.3466,
+      "step": 616
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0239996767201123e-05,
+      "loss": 0.3568,
+      "step": 617
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0213334758848817e-05,
+      "loss": 0.3479,
+      "step": 618
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0186671233199759e-05,
+      "loss": 0.3581,
+      "step": 619
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0160006379892436e-05,
+      "loss": 0.3229,
+      "step": 620
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0133340388574775e-05,
+      "loss": 0.385,
+      "step": 621
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0106673448902802e-05,
+      "loss": 0.3452,
+      "step": 622
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0080005750539288e-05,
+      "loss": 0.3423,
+      "step": 623
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.005333748315239e-05,
+      "loss": 0.3518,
+      "step": 624
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0026668836414325e-05,
+      "loss": 0.3641,
+      "step": 625
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1e-05,
+      "loss": 0.3489,
+      "step": 626
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.97333116358568e-06,
+      "loss": 0.3662,
+      "step": 627
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.946662516847612e-06,
+      "loss": 0.3837,
+      "step": 628
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.919994249460719e-06,
+      "loss": 0.3343,
+      "step": 629
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.893326551097198e-06,
+      "loss": 0.3562,
+      "step": 630
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.866659611425227e-06,
+      "loss": 0.3711,
+      "step": 631
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.839993620107566e-06,
+      "loss": 0.355,
+      "step": 632
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.813328766800243e-06,
+      "loss": 0.361,
+      "step": 633
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.786665241151185e-06,
+      "loss": 0.347,
+      "step": 634
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.760003232798879e-06,
+      "loss": 0.3657,
+      "step": 635
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.733342931371024e-06,
+      "loss": 0.3769,
+      "step": 636
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.706684526483168e-06,
+      "loss": 0.3688,
+      "step": 637
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.680028207737384e-06,
+      "loss": 0.3979,
+      "step": 638
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.653374164720898e-06,
+      "loss": 0.3764,
+      "step": 639
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.62672258700476e-06,
+      "loss": 0.3239,
+      "step": 640
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.600073664142473e-06,
+      "loss": 0.3469,
+      "step": 641
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.573427585668666e-06,
+      "loss": 0.3554,
+      "step": 642
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.54678454109774e-06,
+      "loss": 0.3717,
+      "step": 643
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.52014471992251e-06,
+      "loss": 0.3211,
+      "step": 644
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.493508311612874e-06,
+      "loss": 0.349,
+      "step": 645
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.46687550561445e-06,
+      "loss": 0.3572,
+      "step": 646
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.440246491347244e-06,
+      "loss": 0.3525,
+      "step": 647
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.413621458204282e-06,
+      "loss": 0.3311,
+      "step": 648
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.387000595550282e-06,
+      "loss": 0.3632,
+      "step": 649
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.360384092720303e-06,
+      "loss": 0.3682,
+      "step": 650
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.333772139018388e-06,
+      "loss": 0.3667,
+      "step": 651
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.307164923716234e-06,
+      "loss": 0.3734,
+      "step": 652
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.28056263605183e-06,
+      "loss": 0.3887,
+      "step": 653
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.253965465228122e-06,
+      "loss": 0.3664,
+      "step": 654
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.227373600411668e-06,
+      "loss": 0.354,
+      "step": 655
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.200787230731275e-06,
+      "loss": 0.3402,
+      "step": 656
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.174206545276678e-06,
+      "loss": 0.3721,
+      "step": 657
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.14763173309718e-06,
+      "loss": 0.3578,
+      "step": 658
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.12106298320032e-06,
+      "loss": 0.3653,
+      "step": 659
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.0945004845505e-06,
+      "loss": 0.3603,
+      "step": 660
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.067944426067688e-06,
+      "loss": 0.3478,
+      "step": 661
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.041394996626028e-06,
+      "loss": 0.3488,
+      "step": 662
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.014852385052519e-06,
+      "loss": 0.3864,
+      "step": 663
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.98831678012568e-06,
+      "loss": 0.3866,
+      "step": 664
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.961788370574184e-06,
+      "loss": 0.3393,
+      "step": 665
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.935267345075541e-06,
+      "loss": 0.3488,
+      "step": 666
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.90875389225473e-06,
+      "loss": 0.3636,
+      "step": 667
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.882248200682882e-06,
+      "loss": 0.3591,
+      "step": 668
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.855750458875924e-06,
+      "loss": 0.3682,
+      "step": 669
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.829260855293237e-06,
+      "loss": 0.3197,
+      "step": 670
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.80277957833633e-06,
+      "loss": 0.3653,
+      "step": 671
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.776306816347484e-06,
+      "loss": 0.3653,
+      "step": 672
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.749842757608423e-06,
+      "loss": 0.3618,
+      "step": 673
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.723387590338965e-06,
+      "loss": 0.3561,
+      "step": 674
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.6969415026957e-06,
+      "loss": 0.3717,
+      "step": 675
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.670504682770632e-06,
+      "loss": 0.369,
+      "step": 676
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.644077318589848e-06,
+      "loss": 0.3423,
+      "step": 677
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.617659598112195e-06,
+      "loss": 0.3642,
+      "step": 678
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.59125170922792e-06,
+      "loss": 0.3377,
+      "step": 679
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.564853839757357e-06,
+      "loss": 0.3364,
+      "step": 680
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.538466177449559e-06,
+      "loss": 0.3405,
+      "step": 681
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.512088909981007e-06,
+      "loss": 0.3695,
+      "step": 682
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.485722224954237e-06,
+      "loss": 0.3608,
+      "step": 683
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.459366309896514e-06,
+      "loss": 0.3783,
+      "step": 684
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.433021352258522e-06,
+      "loss": 0.3432,
+      "step": 685
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.406687539412995e-06,
+      "loss": 0.3673,
+      "step": 686
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.380365058653416e-06,
+      "loss": 0.3907,
+      "step": 687
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.35405409719266e-06,
+      "loss": 0.365,
+      "step": 688
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.327754842161685e-06,
+      "loss": 0.33,
+      "step": 689
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.301467480608176e-06,
+      "loss": 0.3311,
+      "step": 690
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.275192199495237e-06,
+      "loss": 0.3706,
+      "step": 691
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.248929185700053e-06,
+      "loss": 0.3506,
+      "step": 692
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.222678626012554e-06,
+      "loss": 0.3219,
+      "step": 693
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.196440707134102e-06,
+      "loss": 0.3456,
+      "step": 694
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.170215615676145e-06,
+      "loss": 0.4205,
+      "step": 695
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.144003538158907e-06,
+      "loss": 0.3794,
+      "step": 696
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.117804661010046e-06,
+      "loss": 0.3444,
+      "step": 697
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.091619170563336e-06,
+      "loss": 0.3715,
+      "step": 698
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.065447253057348e-06,
+      "loss": 0.3767,
+      "step": 699
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.03928909463411e-06,
+      "loss": 0.361,
+      "step": 700
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.013144881337797e-06,
+      "loss": 0.3755,
+      "step": 701
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.987014799113398e-06,
+      "loss": 0.3827,
+      "step": 702
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.960899033805408e-06,
+      "loss": 0.331,
+      "step": 703
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.934797771156482e-06,
+      "loss": 0.3423,
+      "step": 704
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.908711196806131e-06,
+      "loss": 0.3584,
+      "step": 705
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.882639496289414e-06,
+      "loss": 0.3497,
+      "step": 706
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.856582855035578e-06,
+      "loss": 0.3475,
+      "step": 707
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.830541458366787e-06,
+      "loss": 0.3654,
+      "step": 708
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.804515491496765e-06,
+      "loss": 0.3495,
+      "step": 709
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.77850513952951e-06,
+      "loss": 0.388,
+      "step": 710
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.75251058745795e-06,
+      "loss": 0.3737,
+      "step": 711
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.726532020162639e-06,
+      "loss": 0.3623,
+      "step": 712
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.700569622410454e-06,
+      "loss": 0.3726,
+      "step": 713
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.67462357885326e-06,
+      "loss": 0.3868,
+      "step": 714
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.648694074026616e-06,
+      "loss": 0.3685,
+      "step": 715
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.622781292348437e-06,
+      "loss": 0.3431,
+      "step": 716
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.596885418117714e-06,
+      "loss": 0.3628,
+      "step": 717
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.571006635513183e-06,
+      "loss": 0.3873,
+      "step": 718
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.545145128592009e-06,
+      "loss": 0.3479,
+      "step": 719
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.5193010812885055e-06,
+      "loss": 0.3731,
+      "step": 720
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.493474677412795e-06,
+      "loss": 0.3931,
+      "step": 721
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.467666100649522e-06,
+      "loss": 0.3957,
+      "step": 722
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.441875534556532e-06,
+      "loss": 0.3694,
+      "step": 723
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.416103162563583e-06,
+      "loss": 0.3389,
+      "step": 724
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.390349167971026e-06,
+      "loss": 0.3625,
+      "step": 725
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.364613733948503e-06,
+      "loss": 0.3679,
+      "step": 726
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.338897043533657e-06,
+      "loss": 0.3565,
+      "step": 727
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.313199279630815e-06,
+      "loss": 0.3858,
+      "step": 728
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.287520625009698e-06,
+      "loss": 0.3918,
+      "step": 729
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.26186126230411e-06,
+      "loss": 0.3862,
+      "step": 730
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.236221374010648e-06,
+      "loss": 0.3606,
+      "step": 731
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.2106011424874075e-06,
+      "loss": 0.3491,
+      "step": 732
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.185000749952667e-06,
+      "loss": 0.4008,
+      "step": 733
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.159420378483619e-06,
+      "loss": 0.3533,
+      "step": 734
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.133860210015049e-06,
+      "loss": 0.3692,
+      "step": 735
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.108320426338063e-06,
+      "loss": 0.3483,
+      "step": 736
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.082801209098775e-06,
+      "loss": 0.3698,
+      "step": 737
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.057302739797026e-06,
+      "loss": 0.3572,
+      "step": 738
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.031825199785103e-06,
+      "loss": 0.3291,
+      "step": 739
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.006368770266421e-06,
+      "loss": 0.41,
+      "step": 740
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.980933632294269e-06,
+      "loss": 0.369,
+      "step": 741
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.955519966770486e-06,
+      "loss": 0.3709,
+      "step": 742
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.93012795444421e-06,
+      "loss": 0.3935,
+      "step": 743
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.904757775910569e-06,
+      "loss": 0.3588,
+      "step": 744
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.879409611609394e-06,
+      "loss": 0.3703,
+      "step": 745
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.854083641823958e-06,
+      "loss": 0.37,
+      "step": 746
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8287800466796715e-06,
+      "loss": 0.3646,
+      "step": 747
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.80349900614282e-06,
+      "loss": 0.383,
+      "step": 748
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7782407000192586e-06,
+      "loss": 0.3704,
+      "step": 749
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7530053079531664e-06,
+      "loss": 0.366,
+      "step": 750
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7277930094257405e-06,
+      "loss": 0.3968,
+      "step": 751
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.702603983753928e-06,
+      "loss": 0.379,
+      "step": 752
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.677438410089163e-06,
+      "loss": 0.3619,
+      "step": 753
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.652296467416074e-06,
+      "loss": 0.3713,
+      "step": 754
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.627178334551227e-06,
+      "loss": 0.3814,
+      "step": 755
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.602084190141835e-06,
+      "loss": 0.3501,
+      "step": 756
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.577014212664509e-06,
+      "loss": 0.3584,
+      "step": 757
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.551968580423974e-06,
+      "loss": 0.3664,
+      "step": 758
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.526947471551799e-06,
+      "loss": 0.3424,
+      "step": 759
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.501951064005147e-06,
+      "loss": 0.3827,
+      "step": 760
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.476979535565486e-06,
+      "loss": 0.3727,
+      "step": 761
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.452033063837354e-06,
+      "loss": 0.3495,
+      "step": 762
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.427111826247057e-06,
+      "loss": 0.3752,
+      "step": 763
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.402216000041446e-06,
+      "loss": 0.3429,
+      "step": 764
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.377345762286633e-06,
+      "loss": 0.3841,
+      "step": 765
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.35250128986673e-06,
+      "loss": 0.3659,
+      "step": 766
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3276827594826184e-06,
+      "loss": 0.3942,
+      "step": 767
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3028903476506484e-06,
+      "loss": 0.3539,
+      "step": 768
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.278124230701428e-06,
+      "loss": 0.3665,
+      "step": 769
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.253384584778534e-06,
+      "loss": 0.3552,
+      "step": 770
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2286715858372894e-06,
+      "loss": 0.3694,
+      "step": 771
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.203985409643478e-06,
+      "loss": 0.3582,
+      "step": 772
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.179326231772123e-06,
+      "loss": 0.3919,
+      "step": 773
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.154694227606235e-06,
+      "loss": 0.3774,
+      "step": 774
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.130089572335536e-06,
+      "loss": 0.3464,
+      "step": 775
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.105512440955259e-06,
+      "loss": 0.3692,
+      "step": 776
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.080963008264862e-06,
+      "loss": 0.3748,
+      "step": 777
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.056441448866817e-06,
+      "loss": 0.3618,
+      "step": 778
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.031947937165335e-06,
+      "loss": 0.343,
+      "step": 779
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.007482647365159e-06,
+      "loss": 0.3801,
+      "step": 780
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.983045753470308e-06,
+      "loss": 0.3353,
+      "step": 781
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.958637429282831e-06,
+      "loss": 0.3682,
+      "step": 782
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.934257848401594e-06,
+      "loss": 0.388,
+      "step": 783
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.909907184221023e-06,
+      "loss": 0.3616,
+      "step": 784
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.885585609929891e-06,
+      "loss": 0.3975,
+      "step": 785
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.861293298510062e-06,
+      "loss": 0.3292,
+      "step": 786
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.837030422735281e-06,
+      "loss": 0.3984,
+      "step": 787
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.812797155169943e-06,
+      "loss": 0.357,
+      "step": 788
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.788593668167854e-06,
+      "loss": 0.3401,
+      "step": 789
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.764420133871016e-06,
+      "loss": 0.3673,
+      "step": 790
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.740276724208397e-06,
+      "loss": 0.4069,
+      "step": 791
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.716163610894709e-06,
+      "loss": 0.3983,
+      "step": 792
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6920809654291945e-06,
+      "loss": 0.3766,
+      "step": 793
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.668028959094386e-06,
+      "loss": 0.3915,
+      "step": 794
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.644007762954926e-06,
+      "loss": 0.375,
+      "step": 795
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.620017547856296e-06,
+      "loss": 0.3737,
+      "step": 796
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5960584844236565e-06,
+      "loss": 0.3728,
+      "step": 797
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.572130743060597e-06,
+      "loss": 0.3709,
+      "step": 798
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.548234493947939e-06,
+      "loss": 0.3482,
+      "step": 799
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5243699070425196e-06,
+      "loss": 0.4044,
+      "step": 800
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.500537152075987e-06,
+      "loss": 0.3441,
+      "step": 801
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.476736398553592e-06,
+      "loss": 0.3495,
+      "step": 802
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.452967815752981e-06,
+      "loss": 0.3704,
+      "step": 803
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.429231572722995e-06,
+      "loss": 0.3517,
+      "step": 804
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.405527838282458e-06,
+      "loss": 0.3546,
+      "step": 805
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.381856781019006e-06,
+      "loss": 0.3863,
+      "step": 806
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.358218569287834e-06,
+      "loss": 0.3725,
+      "step": 807
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3346133712105496e-06,
+      "loss": 0.3989,
+      "step": 808
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.311041354673965e-06,
+      "loss": 0.4024,
+      "step": 809
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.287502687328869e-06,
+      "loss": 0.3603,
+      "step": 810
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.2639975365888915e-06,
+      "loss": 0.3583,
+      "step": 811
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.240526069629265e-06,
+      "loss": 0.3633,
+      "step": 812
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.217088453385658e-06,
+      "loss": 0.3567,
+      "step": 813
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.193684854552983e-06,
+      "loss": 0.39,
+      "step": 814
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.170315439584212e-06,
+      "loss": 0.3537,
+      "step": 815
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.146980374689192e-06,
+      "loss": 0.3735,
+      "step": 816
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.123679825833459e-06,
+      "loss": 0.3407,
+      "step": 817
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.100413958737067e-06,
+      "loss": 0.3528,
+      "step": 818
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.077182938873394e-06,
+      "loss": 0.3588,
+      "step": 819
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.0539869314679945e-06,
+      "loss": 0.3632,
+      "step": 820
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.03082610149738e-06,
+      "loss": 0.3489,
+      "step": 821
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.0077006136878805e-06,
+      "loss": 0.3742,
+      "step": 822
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.984610632514476e-06,
+      "loss": 0.3353,
+      "step": 823
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.961556322199585e-06,
+      "loss": 0.3434,
+      "step": 824
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.938537846711953e-06,
+      "loss": 0.3774,
+      "step": 825
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.91555536976544e-06,
+      "loss": 0.3856,
+      "step": 826
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.8926090548178836e-06,
+      "loss": 0.3378,
+      "step": 827
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.869699065069921e-06,
+      "loss": 0.3234,
+      "step": 828
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.846825563463838e-06,
+      "loss": 0.3454,
+      "step": 829
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.823988712682406e-06,
+      "loss": 0.3141,
+      "step": 830
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.8011886751477196e-06,
+      "loss": 0.339,
+      "step": 831
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.7784256130200665e-06,
+      "loss": 0.3489,
+      "step": 832
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.755699688196724e-06,
+      "loss": 0.3333,
+      "step": 833
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.733011062310867e-06,
+      "loss": 0.3209,
+      "step": 834
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.710359896730379e-06,
+      "loss": 0.3186,
+      "step": 835
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.6877463525567035e-06,
+      "loss": 0.2997,
+      "step": 836
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.665170590623731e-06,
+      "loss": 0.328,
+      "step": 837
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.642632771496622e-06,
+      "loss": 0.3448,
+      "step": 838
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.620133055470678e-06,
+      "loss": 0.3354,
+      "step": 839
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.597671602570204e-06,
+      "loss": 0.3186,
+      "step": 840
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.575248572547363e-06,
+      "loss": 0.3268,
+      "step": 841
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5528641248810455e-06,
+      "loss": 0.3172,
+      "step": 842
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.530518418775734e-06,
+      "loss": 0.3217,
+      "step": 843
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.508211613160368e-06,
+      "loss": 0.3127,
+      "step": 844
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.485943866687217e-06,
+      "loss": 0.3264,
+      "step": 845
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.463715337730759e-06,
+      "loss": 0.335,
+      "step": 846
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4415261843865256e-06,
+      "loss": 0.3239,
+      "step": 847
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.419376564470019e-06,
+      "loss": 0.2964,
+      "step": 848
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.397266635515563e-06,
+      "loss": 0.3165,
+      "step": 849
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.375196554775174e-06,
+      "loss": 0.319,
+      "step": 850
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3531664792174776e-06,
+      "loss": 0.3071,
+      "step": 851
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.331176565526558e-06,
+      "loss": 0.3027,
+      "step": 852
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.309226970100862e-06,
+      "loss": 0.304,
+      "step": 853
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.287317849052075e-06,
+      "loss": 0.2889,
+      "step": 854
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2654493582040345e-06,
+      "loss": 0.3014,
+      "step": 855
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.243621653091581e-06,
+      "loss": 0.2998,
+      "step": 856
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.221834888959487e-06,
+      "loss": 0.2954,
+      "step": 857
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.200089220761353e-06,
+      "loss": 0.3216,
+      "step": 858
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.178384803158465e-06,
+      "loss": 0.3117,
+      "step": 859
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.156721790518754e-06,
+      "loss": 0.3121,
+      "step": 860
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.135100336915648e-06,
+      "loss": 0.3401,
+      "step": 861
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.1135205961269985e-06,
+      "loss": 0.3192,
+      "step": 862
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.091982721633989e-06,
+      "loss": 0.3141,
+      "step": 863
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.070486866620035e-06,
+      "loss": 0.306,
+      "step": 864
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.049033183969697e-06,
+      "loss": 0.2947,
+      "step": 865
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.027621826267593e-06,
+      "loss": 0.2918,
+      "step": 866
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.00625294579732e-06,
+      "loss": 0.3051,
+      "step": 867
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.984926694540352e-06,
+      "loss": 0.3103,
+      "step": 868
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.963643224174994e-06,
+      "loss": 0.3108,
+      "step": 869
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.942402686075258e-06,
+      "loss": 0.3056,
+      "step": 870
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.921205231309818e-06,
+      "loss": 0.2914,
+      "step": 871
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.90005101064094e-06,
+      "loss": 0.2873,
+      "step": 872
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.878940174523371e-06,
+      "loss": 0.2775,
+      "step": 873
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.857872873103322e-06,
+      "loss": 0.3411,
+      "step": 874
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.836849256217356e-06,
+      "loss": 0.2879,
+      "step": 875
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8158694733913435e-06,
+      "loss": 0.3133,
+      "step": 876
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.794933673839396e-06,
+      "loss": 0.3019,
+      "step": 877
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7740420064628035e-06,
+      "loss": 0.2722,
+      "step": 878
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.753194619848973e-06,
+      "loss": 0.3058,
+      "step": 879
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.732391662270376e-06,
+      "loss": 0.2969,
+      "step": 880
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7116332816834998e-06,
+      "loss": 0.2978,
+      "step": 881
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6909196257277678e-06,
+      "loss": 0.2663,
+      "step": 882
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.670250841724533e-06,
+      "loss": 0.3129,
+      "step": 883
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6496270766759934e-06,
+      "loss": 0.2908,
+      "step": 884
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.629048477264152e-06,
+      "loss": 0.3289,
+      "step": 885
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6085151898498004e-06,
+      "loss": 0.3035,
+      "step": 886
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5880273604714466e-06,
+      "loss": 0.3037,
+      "step": 887
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.567585134844288e-06,
+      "loss": 0.2735,
+      "step": 888
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5471886583591795e-06,
+      "loss": 0.2811,
+      "step": 889
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5268380760815925e-06,
+      "loss": 0.3095,
+      "step": 890
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5065335327505867e-06,
+      "loss": 0.284,
+      "step": 891
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.48627517277778e-06,
+      "loss": 0.329,
+      "step": 892
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.466063140246321e-06,
+      "loss": 0.2915,
+      "step": 893
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4458975789098614e-06,
+      "loss": 0.309,
+      "step": 894
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.425778632191551e-06,
+      "loss": 0.2887,
+      "step": 895
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4057064431829765e-06,
+      "loss": 0.2915,
+      "step": 896
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3856811546432e-06,
+      "loss": 0.2823,
+      "step": 897
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3657029089976987e-06,
+      "loss": 0.3256,
+      "step": 898
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.345771848337359e-06,
+      "loss": 0.3282,
+      "step": 899
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.325888114417497e-06,
+      "loss": 0.3075,
+      "step": 900
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3060518486568103e-06,
+      "loss": 0.3166,
+      "step": 901
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.2862631921363965e-06,
+      "loss": 0.2931,
+      "step": 902
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.26652228559874e-06,
+      "loss": 0.2983,
+      "step": 903
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2468292694467165e-06,
+      "loss": 0.3031,
+      "step": 904
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2271842837425917e-06,
+      "loss": 0.308,
+      "step": 905
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2075874682070185e-06,
+      "loss": 0.3048,
+      "step": 906
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1880389622180664e-06,
+      "loss": 0.3295,
+      "step": 907
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.168538904810191e-06,
+      "loss": 0.3288,
+      "step": 908
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1490874346732904e-06,
+      "loss": 0.3148,
+      "step": 909
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1296846901516808e-06,
+      "loss": 0.3255,
+      "step": 910
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.110330809243134e-06,
+      "loss": 0.3064,
+      "step": 911
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.091025929597892e-06,
+      "loss": 0.3341,
+      "step": 912
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0717701885176797e-06,
+      "loss": 0.2954,
+      "step": 913
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.052563722954741e-06,
+      "loss": 0.2878,
+      "step": 914
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0334066695108566e-06,
+      "loss": 0.3033,
+      "step": 915
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.014299164436372e-06,
+      "loss": 0.3326,
+      "step": 916
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9952413436292306e-06,
+      "loss": 0.3276,
+      "step": 917
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9762333426340173e-06,
+      "loss": 0.3063,
+      "step": 918
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.957275296640969e-06,
+      "loss": 0.295,
+      "step": 919
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9383673404850354e-06,
+      "loss": 0.2937,
+      "step": 920
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9195096086449227e-06,
+      "loss": 0.2934,
+      "step": 921
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9007022352421068e-06,
+      "loss": 0.3254,
+      "step": 922
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.881945354039921e-06,
+      "loss": 0.2959,
+      "step": 923
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8632390984425753e-06,
+      "loss": 0.3119,
+      "step": 924
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8445836014942073e-06,
+      "loss": 0.3412,
+      "step": 925
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.825978995877964e-06,
+      "loss": 0.3323,
+      "step": 926
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8074254139150223e-06,
+      "loss": 0.3183,
+      "step": 927
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7889229875636726e-06,
+      "loss": 0.3281,
+      "step": 928
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7704718484183703e-06,
+      "loss": 0.3156,
+      "step": 929
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7520721277088023e-06,
+      "loss": 0.2926,
+      "step": 930
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7337239562989517e-06,
+      "loss": 0.296,
+      "step": 931
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.715427464686169e-06,
+      "loss": 0.3085,
+      "step": 932
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.697182783000246e-06,
+      "loss": 0.328,
+      "step": 933
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6789900410024804e-06,
+      "loss": 0.3022,
+      "step": 934
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6608493680847757e-06,
+      "loss": 0.3138,
+      "step": 935
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.642760893268684e-06,
+      "loss": 0.316,
+      "step": 936
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6247247452045287e-06,
+      "loss": 0.3229,
+      "step": 937
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6067410521704597e-06,
+      "loss": 0.2877,
+      "step": 938
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5888099420715573e-06,
+      "loss": 0.3112,
+      "step": 939
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5709315424389135e-06,
+      "loss": 0.3344,
+      "step": 940
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5531059804287306e-06,
+      "loss": 0.3032,
+      "step": 941
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5353333828214156e-06,
+      "loss": 0.2967,
+      "step": 942
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5176138760206737e-06,
+      "loss": 0.2873,
+      "step": 943
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4999475860526236e-06,
+      "loss": 0.3084,
+      "step": 944
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4823346385648774e-06,
+      "loss": 0.3254,
+      "step": 945
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4647751588256653e-06,
+      "loss": 0.2906,
+      "step": 946
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4472692717229505e-06,
+      "loss": 0.3372,
+      "step": 947
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.429817101763511e-06,
+      "loss": 0.3175,
+      "step": 948
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4124187730720916e-06,
+      "loss": 0.3344,
+      "step": 949
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.395074409390491e-06,
+      "loss": 0.305,
+      "step": 950
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.377784134076698e-06,
+      "loss": 0.354,
+      "step": 951
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3605480701040095e-06,
+      "loss": 0.3228,
+      "step": 952
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.343366340060157e-06,
+      "loss": 0.3064,
+      "step": 953
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3262390661464307e-06,
+      "loss": 0.3194,
+      "step": 954
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3091663701768165e-06,
+      "loss": 0.3052,
+      "step": 955
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2921483735771255e-06,
+      "loss": 0.3289,
+      "step": 956
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.275185197384129e-06,
+      "loss": 0.3102,
+      "step": 957
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.258276962244711e-06,
+      "loss": 0.3502,
+      "step": 958
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2414237884149823e-06,
+      "loss": 0.3402,
+      "step": 959
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.224625795759451e-06,
+      "loss": 0.326,
+      "step": 960
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.207883103750171e-06,
+      "loss": 0.3168,
+      "step": 961
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.1911958314658598e-06,
+      "loss": 0.3301,
+      "step": 962
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.1745640975910964e-06,
+      "loss": 0.3119,
+      "step": 963
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1579880204154414e-06,
+      "loss": 0.3536,
+      "step": 964
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.141467717832616e-06,
+      "loss": 0.2937,
+      "step": 965
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1250033073396524e-06,
+      "loss": 0.3094,
+      "step": 966
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1085949060360654e-06,
+      "loss": 0.333,
+      "step": 967
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.092242630623016e-06,
+      "loss": 0.3111,
+      "step": 968
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0759465974024786e-06,
+      "loss": 0.3226,
+      "step": 969
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.05970692227643e-06,
+      "loss": 0.3075,
+      "step": 970
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.04352372074599e-06,
+      "loss": 0.3209,
+      "step": 971
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.027397107910647e-06,
+      "loss": 0.3221,
+      "step": 972
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0113271984674e-06,
+      "loss": 0.3122,
+      "step": 973
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9953141067099537e-06,
+      "loss": 0.3082,
+      "step": 974
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9793579465279245e-06,
+      "loss": 0.3206,
+      "step": 975
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9634588314060055e-06,
+      "loss": 0.3185,
+      "step": 976
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9476168744231694e-06,
+      "loss": 0.3118,
+      "step": 977
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9318321882518675e-06,
+      "loss": 0.3334,
+      "step": 978
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9161048851572216e-06,
+      "loss": 0.3391,
+      "step": 979
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9004350769962332e-06,
+      "loss": 0.3201,
+      "step": 980
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8848228752169796e-06,
+      "loss": 0.3331,
+      "step": 981
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8692683908578268e-06,
+      "loss": 0.2984,
+      "step": 982
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8537717345466356e-06,
+      "loss": 0.3065,
+      "step": 983
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8383330164999902e-06,
+      "loss": 0.3169,
+      "step": 984
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8229523465223787e-06,
+      "loss": 0.3165,
+      "step": 985
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8076298340054565e-06,
+      "loss": 0.3234,
+      "step": 986
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.7923655879272395e-06,
+      "loss": 0.2998,
+      "step": 987
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7771597168513267e-06,
+      "loss": 0.3244,
+      "step": 988
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7620123289261527e-06,
+      "loss": 0.3333,
+      "step": 989
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.746923531884196e-06,
+      "loss": 0.3164,
+      "step": 990
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7318934330412197e-06,
+      "loss": 0.3309,
+      "step": 991
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7169221392955092e-06,
+      "loss": 0.334,
+      "step": 992
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7020097571271188e-06,
+      "loss": 0.3508,
+      "step": 993
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6871563925970946e-06,
+      "loss": 0.3293,
+      "step": 994
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6723621513467381e-06,
+      "loss": 0.3129,
+      "step": 995
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.657627138596858e-06,
+      "loss": 0.3209,
+      "step": 996
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6429514591469953e-06,
+      "loss": 0.3015,
+      "step": 997
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6283352173747148e-06,
+      "loss": 0.3136,
+      "step": 998
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6137785172348308e-06,
+      "loss": 0.3169,
+      "step": 999
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5992814622586872e-06,
+      "loss": 0.3115,
+      "step": 1000
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5848441555534111e-06,
+      "loss": 0.3412,
+      "step": 1001
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5704666998011852e-06,
+      "loss": 0.3018,
+      "step": 1002
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5561491972585152e-06,
+      "loss": 0.3297,
+      "step": 1003
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.541891749755503e-06,
+      "loss": 0.3364,
+      "step": 1004
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5276944586951204e-06,
+      "loss": 0.3317,
+      "step": 1005
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5135574250524898e-06,
+      "loss": 0.3319,
+      "step": 1006
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.4994807493741725e-06,
+      "loss": 0.3247,
+      "step": 1007
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4854645317774363e-06,
+      "loss": 0.3376,
+      "step": 1008
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4715088719495573e-06,
+      "loss": 0.3341,
+      "step": 1009
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4576138691471187e-06,
+      "loss": 0.3052,
+      "step": 1010
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4437796221952748e-06,
+      "loss": 0.3382,
+      "step": 1011
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4300062294870843e-06,
+      "loss": 0.3156,
+      "step": 1012
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4162937889827833e-06,
+      "loss": 0.3337,
+      "step": 1013
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4026423982091043e-06,
+      "loss": 0.3362,
+      "step": 1014
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3890521542585723e-06,
+      "loss": 0.314,
+      "step": 1015
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3755231537888225e-06,
+      "loss": 0.3187,
+      "step": 1016
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3620554930219077e-06,
+      "loss": 0.3571,
+      "step": 1017
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3486492677436125e-06,
+      "loss": 0.3392,
+      "step": 1018
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3353045733027858e-06,
+      "loss": 0.3294,
+      "step": 1019
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3220215046106355e-06,
+      "loss": 0.3326,
+      "step": 1020
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3088001561400853e-06,
+      "loss": 0.293,
+      "step": 1021
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2956406219250817e-06,
+      "loss": 0.3323,
+      "step": 1022
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2825429955599212e-06,
+      "loss": 0.3253,
+      "step": 1023
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2695073701986105e-06,
+      "loss": 0.3253,
+      "step": 1024
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2565338385541792e-06,
+      "loss": 0.3105,
+      "step": 1025
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2436224928980278e-06,
+      "loss": 0.3269,
+      "step": 1026
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2307734250592773e-06,
+      "loss": 0.3375,
+      "step": 1027
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2179867264241064e-06,
+      "loss": 0.3162,
+      "step": 1028
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2052624879351105e-06,
+      "loss": 0.3349,
+      "step": 1029
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1926008000906486e-06,
+      "loss": 0.3341,
+      "step": 1030
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.180001752944202e-06,
+      "loss": 0.3159,
+      "step": 1031
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.167465436103733e-06,
+      "loss": 0.3444,
+      "step": 1032
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1549919387310571e-06,
+      "loss": 0.335,
+      "step": 1033
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.142581349541182e-06,
+      "loss": 0.3389,
+      "step": 1034
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1302337568017142e-06,
+      "loss": 0.3124,
+      "step": 1035
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1179492483322007e-06,
+      "loss": 0.3306,
+      "step": 1036
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1057279115035124e-06,
+      "loss": 0.3469,
+      "step": 1037
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.093569833237238e-06,
+      "loss": 0.3341,
+      "step": 1038
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.081475100005044e-06,
+      "loss": 0.3241,
+      "step": 1039
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0694437978280702e-06,
+      "loss": 0.3429,
+      "step": 1040
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0574760122763217e-06,
+      "loss": 0.3383,
+      "step": 1041
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0455718284680506e-06,
+      "loss": 0.368,
+      "step": 1042
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0337313310691566e-06,
+      "loss": 0.3428,
+      "step": 1043
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0219546042925842e-06,
+      "loss": 0.3703,
+      "step": 1044
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0102417318977253e-06,
+      "loss": 0.326,
+      "step": 1045
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.985927971898178e-07,
+      "loss": 0.3061,
+      "step": 1046
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.870078830193631e-07,
+      "loss": 0.3246,
+      "step": 1047
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.75487071781518e-07,
+      "loss": 0.3276,
+      "step": 1048
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.64030445415537e-07,
+      "loss": 0.3225,
+      "step": 1049
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.526380854041639e-07,
+      "loss": 0.3057,
+      "step": 1050
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.413100727730628e-07,
+      "loss": 0.3396,
+      "step": 1051
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.300464880902449e-07,
+      "loss": 0.3161,
+      "step": 1052
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.188474114654877e-07,
+      "loss": 0.3464,
+      "step": 1053
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.077129225497727e-07,
+      "loss": 0.3033,
+      "step": 1054
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.966431005347109e-07,
+      "loss": 0.3555,
+      "step": 1055
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.856380241519935e-07,
+      "loss": 0.3343,
+      "step": 1056
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.746977716728101e-07,
+      "loss": 0.3307,
+      "step": 1057
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.638224209073098e-07,
+      "loss": 0.3746,
+      "step": 1058
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.530120492040506e-07,
+      "loss": 0.3472,
+      "step": 1059
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.42266733449425e-07,
+      "loss": 0.3382,
+      "step": 1060
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.31586550067145e-07,
+      "loss": 0.3224,
+      "step": 1061
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.209715750176728e-07,
+      "loss": 0.3471,
+      "step": 1062
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.10421883797694e-07,
+      "loss": 0.3278,
+      "step": 1063
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 7.999375514395779e-07,
+      "loss": 0.3534,
+      "step": 1064
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.895186525108411e-07,
+      "loss": 0.3413,
+      "step": 1065
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.791652611136214e-07,
+      "loss": 0.3205,
+      "step": 1066
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.688774508841479e-07,
+      "loss": 0.3237,
+      "step": 1067
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.586552949922176e-07,
+      "loss": 0.3549,
+      "step": 1068
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.484988661406734e-07,
+      "loss": 0.3514,
+      "step": 1069
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.38408236564897e-07,
+      "loss": 0.3394,
+      "step": 1070
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.283834780322763e-07,
+      "loss": 0.339,
+      "step": 1071
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.184246618417112e-07,
+      "loss": 0.3218,
+      "step": 1072
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.085318588231049e-07,
+      "loss": 0.3111,
+      "step": 1073
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.987051393368471e-07,
+      "loss": 0.3619,
+      "step": 1074
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.889445732733324e-07,
+      "loss": 0.3235,
+      "step": 1075
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.792502300524473e-07,
+      "loss": 0.3039,
+      "step": 1076
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.696221786230861e-07,
+      "loss": 0.3773,
+      "step": 1077
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.600604874626549e-07,
+      "loss": 0.3399,
+      "step": 1078
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.505652245765881e-07,
+      "loss": 0.339,
+      "step": 1079
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.411364574978651e-07,
+      "loss": 0.3418,
+      "step": 1080
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.317742532865234e-07,
+      "loss": 0.3324,
+      "step": 1081
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.22478678529197e-07,
+      "loss": 0.3489,
+      "step": 1082
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.132497993386166e-07,
+      "loss": 0.3338,
+      "step": 1083
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.040876813531715e-07,
+      "loss": 0.344,
+      "step": 1084
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.949923897364174e-07,
+      "loss": 0.3064,
+      "step": 1085
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.859639891766211e-07,
+      "loss": 0.3294,
+      "step": 1086
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.77002543886308e-07,
+      "loss": 0.3161,
+      "step": 1087
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.681081176017944e-07,
+      "loss": 0.3578,
+      "step": 1088
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.592807735827399e-07,
+      "loss": 0.3705,
+      "step": 1089
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.505205746116937e-07,
+      "loss": 0.326,
+      "step": 1090
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.418275829936537e-07,
+      "loss": 0.3487,
+      "step": 1091
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.332018605556189e-07,
+      "loss": 0.357,
+      "step": 1092
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.246434686461521e-07,
+      "loss": 0.3549,
+      "step": 1093
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.161524681349394e-07,
+      "loss": 0.326,
+      "step": 1094
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.077289194123624e-07,
+      "loss": 0.3089,
+      "step": 1095
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.99372882389072e-07,
+      "loss": 0.3458,
+      "step": 1096
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.91084416495543e-07,
+      "loss": 0.3019,
+      "step": 1097
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.828635806816806e-07,
+      "loss": 0.3326,
+      "step": 1098
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.7471043341637877e-07,
+      "loss": 0.3273,
+      "step": 1099
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.666250326871069e-07,
+      "loss": 0.3779,
+      "step": 1100
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5860743599951186e-07,
+      "loss": 0.3536,
+      "step": 1101
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.506577003769919e-07,
+      "loss": 0.3434,
+      "step": 1102
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4277588236030234e-07,
+      "loss": 0.3455,
+      "step": 1103
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.349620380071451e-07,
+      "loss": 0.3577,
+      "step": 1104
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.272162228917809e-07,
+      "loss": 0.3339,
+      "step": 1105
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1953849210462084e-07,
+      "loss": 0.3541,
+      "step": 1106
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.119289002518423e-07,
+      "loss": 0.3577,
+      "step": 1107
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.0438750145500474e-07,
+      "loss": 0.3526,
+      "step": 1108
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.9691434935065107e-07,
+      "loss": 0.2979,
+      "step": 1109
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.895094970899427e-07,
+      "loss": 0.3456,
+      "step": 1110
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8217299733826817e-07,
+      "loss": 0.348,
+      "step": 1111
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.749049022748763e-07,
+      "loss": 0.3059,
+      "step": 1112
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.6770526359250047e-07,
+      "loss": 0.3606,
+      "step": 1113
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.605741324969936e-07,
+      "loss": 0.3462,
+      "step": 1114
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5351155970696603e-07,
+      "loss": 0.3412,
+      "step": 1115
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.4651759545341834e-07,
+      "loss": 0.3489,
+      "step": 1116
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.395922894793891e-07,
+      "loss": 0.3693,
+      "step": 1117
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.327356910396018e-07,
+      "loss": 0.359,
+      "step": 1118
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.259478489001111e-07,
+      "loss": 0.3756,
+      "step": 1119
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.1922881133795827e-07,
+      "loss": 0.3569,
+      "step": 1120
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.1257862614082257e-07,
+      "loss": 0.3166,
+      "step": 1121
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.059973406066963e-07,
+      "loss": 0.3383,
+      "step": 1122
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9948500154352687e-07,
+      "loss": 0.3632,
+      "step": 1123
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9304165526890083e-07,
+      "loss": 0.3463,
+      "step": 1124
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.8666734760970927e-07,
+      "loss": 0.3479,
+      "step": 1125
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8036212390181504e-07,
+      "loss": 0.3425,
+      "step": 1126
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7412602898974516e-07,
+      "loss": 0.4172,
+      "step": 1127
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.6795910722635766e-07,
+      "loss": 0.3668,
+      "step": 1128
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.61861402472533e-07,
+      "loss": 0.3342,
+      "step": 1129
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5583295809685993e-07,
+      "loss": 0.3419,
+      "step": 1130
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.498738169753323e-07,
+      "loss": 0.347,
+      "step": 1131
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4398402149103164e-07,
+      "loss": 0.3448,
+      "step": 1132
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.3816361353384054e-07,
+      "loss": 0.3512,
+      "step": 1133
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3241263450013762e-07,
+      "loss": 0.3842,
+      "step": 1134
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2673112529249752e-07,
+      "loss": 0.3695,
+      "step": 1135
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.211191263194157e-07,
+      "loss": 0.3607,
+      "step": 1136
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.155766774950019e-07,
+      "loss": 0.3467,
+      "step": 1137
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1010381823871052e-07,
+      "loss": 0.3334,
+      "step": 1138
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0470058747505516e-07,
+      "loss": 0.3871,
+      "step": 1139
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9936702363333116e-07,
+      "loss": 0.352,
+      "step": 1140
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9410316464734235e-07,
+      "loss": 0.3222,
+      "step": 1141
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8890904795513476e-07,
+      "loss": 0.3487,
+      "step": 1142
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8378471049872448e-07,
+      "loss": 0.3682,
+      "step": 1143
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.7873018872383797e-07,
+      "loss": 0.3273,
+      "step": 1144
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.7374551857965882e-07,
+      "loss": 0.3451,
+      "step": 1145
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.688307355185592e-07,
+      "loss": 0.3873,
+      "step": 1146
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6398587449585557e-07,
+      "loss": 0.3353,
+      "step": 1147
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.592109699695643e-07,
+      "loss": 0.3893,
+      "step": 1148
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.5450605590014544e-07,
+      "loss": 0.3528,
+      "step": 1149
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.498711657502716e-07,
+      "loss": 0.3512,
+      "step": 1150
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.453063324845827e-07,
+      "loss": 0.3616,
+      "step": 1151
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.408115885694572e-07,
+      "loss": 0.3468,
+      "step": 1152
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.3638696597277678e-07,
+      "loss": 0.3595,
+      "step": 1153
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3203249616369873e-07,
+      "loss": 0.3728,
+      "step": 1154
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.277482101124383e-07,
+      "loss": 0.3661,
+      "step": 1155
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.23534138290039e-07,
+      "loss": 0.3559,
+      "step": 1156
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.193903106681671e-07,
+      "loss": 0.3751,
+      "step": 1157
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1531675671888621e-07,
+      "loss": 0.3554,
+      "step": 1158
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1131350541445873e-07,
+      "loss": 0.3533,
+      "step": 1159
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0738058522713147e-07,
+      "loss": 0.3724,
+      "step": 1160
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0351802412893797e-07,
+      "loss": 0.3425,
+      "step": 1161
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.972584959149989e-08,
+      "loss": 0.3448,
+      "step": 1162
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.60040885858271e-08,
+      "loss": 0.3514,
+      "step": 1163
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.235276758212896e-08,
+      "loss": 0.3654,
+      "step": 1164
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.87719125496278e-08,
+      "loss": 0.3442,
+      "step": 1165
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.526154895636907e-08,
+      "loss": 0.3437,
+      "step": 1166
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.182170176904702e-08,
+      "loss": 0.3738,
+      "step": 1167
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.845239545282046e-08,
+      "loss": 0.3404,
+      "step": 1168
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.515365397114282e-08,
+      "loss": 0.3435,
+      "step": 1169
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.192550078559013e-08,
+      "loss": 0.3629,
+      "step": 1170
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.876795885569554e-08,
+      "loss": 0.3609,
+      "step": 1171
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.568105063878394e-08,
+      "loss": 0.3862,
+      "step": 1172
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.26647980898143e-08,
+      "loss": 0.3496,
+      "step": 1173
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.97192226612242e-08,
+      "loss": 0.3379,
+      "step": 1174
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.684434530277005e-08,
+      "loss": 0.3665,
+      "step": 1175
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.404018646138931e-08,
+      "loss": 0.339,
+      "step": 1176
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.1306766081048456e-08,
+      "loss": 0.3615,
+      "step": 1177
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.8644103602600856e-08,
+      "loss": 0.3472,
+      "step": 1178
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.605221796365023e-08,
+      "loss": 0.3756,
+      "step": 1179
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.3531127598414045e-08,
+      "loss": 0.3553,
+      "step": 1180
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.108085043759813e-08,
+      "loss": 0.3528,
+      "step": 1181
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.870140390826005e-08,
+      "loss": 0.3515,
+      "step": 1182
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.639280493369368e-08,
+      "loss": 0.3643,
+      "step": 1183
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.4155069933301535e-08,
+      "loss": 0.3604,
+      "step": 1184
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.198821482248593e-08,
+      "loss": 0.3396,
+      "step": 1185
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.9892255012528013e-08,
+      "loss": 0.3432,
+      "step": 1186
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.7867205410484488e-08,
+      "loss": 0.3412,
+      "step": 1187
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.5913080419075477e-08,
+      "loss": 0.3991,
+      "step": 1188
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.4029893936586835e-08,
+      "loss": 0.3366,
+      "step": 1189
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.2217659356771337e-08,
+      "loss": 0.372,
+      "step": 1190
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0476389568749867e-08,
+      "loss": 0.3649,
+      "step": 1191
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8806096956923704e-08,
+      "loss": 0.3541,
+      "step": 1192
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.7206793400880163e-08,
+      "loss": 0.369,
+      "step": 1193
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.5678490275319315e-08,
+      "loss": 0.3615,
+      "step": 1194
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4221198449960727e-08,
+      "loss": 0.3421,
+      "step": 1195
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2834928289472415e-08,
+      "loss": 0.3645,
+      "step": 1196
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.151968965339756e-08,
+      "loss": 0.409,
+      "step": 1197
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.0275491896084566e-08,
+      "loss": 0.3613,
+      "step": 1198
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.102343866616014e-09,
+      "loss": 0.3601,
+      "step": 1199
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.000253908752031e-09,
+      "loss": 0.3887,
+      "step": 1200
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.96922986086368e-09,
+      "loss": 0.3634,
+      "step": 1201
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.0092790558852195e-09,
+      "loss": 0.3666,
+      "step": 1202
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.120408321256376e-09,
+      "loss": 0.3657,
+      "step": 1203
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.302623978875709e-09,
+      "loss": 0.3601,
+      "step": 1204
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.555931845053984e-09,
+      "loss": 0.3597,
+      "step": 1205
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.8803372304775366e-09,
+      "loss": 0.3701,
+      "step": 1206
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.2758449401638628e-09,
+      "loss": 0.3322,
+      "step": 1207
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.7424592734316403e-09,
+      "loss": 0.3611,
+      "step": 1208
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.2801840238707565e-09,
+      "loss": 0.3552,
+      "step": 1209
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.890224793123292e-10,
+      "loss": 0.3325,
+      "step": 1210
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.689774218065047e-10,
+      "loss": 0.3778,
+      "step": 1211
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.2005112760358225e-10,
+      "loss": 0.3701,
+      "step": 1212
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4224536713847158e-10,
+      "loss": 0.3853,
+      "step": 1213
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.556140501514982e-11,
+      "loss": 0.3782,
+      "step": 1214
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0,
+      "loss": 0.3822,
+      "step": 1215
+    },
+    {
+      "epoch": 3.0,
+      "step": 1215,
+      "total_flos": 3.10921308999254e+18,
+      "train_loss": 0.3849627566926273,
+      "train_runtime": 41204.2102,
+      "train_samples_per_second": 3.779,
+      "train_steps_per_second": 0.029
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 1215,
+  "num_train_epochs": 3,
+  "save_steps": 500,
+  "total_flos": 3.10921308999254e+18,
+  "trial_name": null,
+  "trial_params": null
+}