diff --git "a/Chinese_tiny_llm-zh_105k_en_105k/trainer_state.json" "b/Chinese_tiny_llm-zh_105k_en_105k/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/Chinese_tiny_llm-zh_105k_en_105k/trainer_state.json"
@@ -0,0 +1,29577 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.9990867579908675,
+  "eval_steps": 500,
+  "global_step": 4926,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.097165991902835e-08,
+      "loss": 2.3529,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.619433198380567e-07,
+      "loss": 1.7259,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4291497975708504e-07,
+      "loss": 1.7718,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.238866396761134e-07,
+      "loss": 1.8515,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.048582995951417e-07,
+      "loss": 1.6216,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.858299595141701e-07,
+      "loss": 1.9882,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.668016194331984e-07,
+      "loss": 1.5576,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.477732793522268e-07,
+      "loss": 1.6755,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.287449392712551e-07,
+      "loss": 1.7665,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.097165991902834e-07,
+      "loss": 1.7105,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.906882591093118e-07,
+      "loss": 1.7106,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.716599190283402e-07,
+      "loss": 1.6979,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0526315789473685e-06,
+      "loss": 1.7277,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.133603238866397e-06,
+      "loss": 1.8052,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.214574898785425e-06,
+      "loss": 1.7795,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2955465587044536e-06,
+      "loss": 1.6371,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3765182186234818e-06,
+      "loss": 1.8089,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4574898785425101e-06,
+      "loss": 1.8317,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5384615384615387e-06,
+      "loss": 1.8051,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6194331983805669e-06,
+      "loss": 2.1316,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7004048582995952e-06,
+      "loss": 1.7696,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7813765182186236e-06,
+      "loss": 1.5114,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.862348178137652e-06,
+      "loss": 1.8426,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9433198380566803e-06,
+      "loss": 1.6995,
+      "step": 24
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0242914979757085e-06,
+      "loss": 1.6744,
+      "step": 25
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.105263157894737e-06,
+      "loss": 1.6003,
+      "step": 26
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1862348178137656e-06,
+      "loss": 1.6063,
+      "step": 27
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.267206477732794e-06,
+      "loss": 1.5973,
+      "step": 28
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.348178137651822e-06,
+      "loss": 1.4834,
+      "step": 29
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.42914979757085e-06,
+      "loss": 1.5489,
+      "step": 30
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5101214574898787e-06,
+      "loss": 1.6397,
+      "step": 31
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5910931174089072e-06,
+      "loss": 1.5741,
+      "step": 32
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.672064777327936e-06,
+      "loss": 1.6861,
+      "step": 33
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7530364372469636e-06,
+      "loss": 1.6109,
+      "step": 34
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.834008097165992e-06,
+      "loss": 1.364,
+      "step": 35
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9149797570850203e-06,
+      "loss": 1.4764,
+      "step": 36
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.995951417004049e-06,
+      "loss": 1.4712,
+      "step": 37
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0769230769230774e-06,
+      "loss": 1.6318,
+      "step": 38
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.157894736842105e-06,
+      "loss": 1.6722,
+      "step": 39
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2388663967611337e-06,
+      "loss": 1.5402,
+      "step": 40
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3198380566801623e-06,
+      "loss": 1.6959,
+      "step": 41
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.4008097165991905e-06,
+      "loss": 1.5145,
+      "step": 42
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.481781376518219e-06,
+      "loss": 1.4529,
+      "step": 43
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.562753036437247e-06,
+      "loss": 1.3066,
+      "step": 44
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.6437246963562758e-06,
+      "loss": 1.5591,
+      "step": 45
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.724696356275304e-06,
+      "loss": 1.5939,
+      "step": 46
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.805668016194332e-06,
+      "loss": 1.6672,
+      "step": 47
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.886639676113361e-06,
+      "loss": 1.4926,
+      "step": 48
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.967611336032389e-06,
+      "loss": 1.5751,
+      "step": 49
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.048582995951417e-06,
+      "loss": 1.6146,
+      "step": 50
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.1295546558704455e-06,
+      "loss": 1.5743,
+      "step": 51
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.210526315789474e-06,
+      "loss": 1.4544,
+      "step": 52
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.291497975708503e-06,
+      "loss": 1.5594,
+      "step": 53
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.372469635627531e-06,
+      "loss": 1.5089,
+      "step": 54
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.453441295546559e-06,
+      "loss": 1.397,
+      "step": 55
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.534412955465588e-06,
+      "loss": 1.3116,
+      "step": 56
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.615384615384616e-06,
+      "loss": 1.6116,
+      "step": 57
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.696356275303644e-06,
+      "loss": 1.3563,
+      "step": 58
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.7773279352226725e-06,
+      "loss": 1.6231,
+      "step": 59
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.8582995951417e-06,
+      "loss": 1.6718,
+      "step": 60
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.939271255060729e-06,
+      "loss": 1.5459,
+      "step": 61
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.020242914979757e-06,
+      "loss": 1.4504,
+      "step": 62
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.101214574898786e-06,
+      "loss": 1.6278,
+      "step": 63
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.1821862348178145e-06,
+      "loss": 1.4232,
+      "step": 64
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.263157894736842e-06,
+      "loss": 1.7075,
+      "step": 65
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.344129554655872e-06,
+      "loss": 1.3666,
+      "step": 66
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.425101214574899e-06,
+      "loss": 1.5153,
+      "step": 67
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.506072874493927e-06,
+      "loss": 1.4662,
+      "step": 68
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.5870445344129565e-06,
+      "loss": 1.4809,
+      "step": 69
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.668016194331984e-06,
+      "loss": 1.3277,
+      "step": 70
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.748987854251013e-06,
+      "loss": 1.5316,
+      "step": 71
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.8299595141700406e-06,
+      "loss": 1.6827,
+      "step": 72
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.91093117408907e-06,
+      "loss": 1.2721,
+      "step": 73
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.991902834008098e-06,
+      "loss": 1.5622,
+      "step": 74
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.0728744939271254e-06,
+      "loss": 1.5434,
+      "step": 75
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.153846153846155e-06,
+      "loss": 1.7226,
+      "step": 76
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.234817813765183e-06,
+      "loss": 1.6995,
+      "step": 77
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.31578947368421e-06,
+      "loss": 1.4116,
+      "step": 78
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.39676113360324e-06,
+      "loss": 1.6227,
+      "step": 79
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.4777327935222675e-06,
+      "loss": 1.6159,
+      "step": 80
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.558704453441296e-06,
+      "loss": 1.4934,
+      "step": 81
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.639676113360325e-06,
+      "loss": 1.5734,
+      "step": 82
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.720647773279353e-06,
+      "loss": 1.6893,
+      "step": 83
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.801619433198381e-06,
+      "loss": 1.4518,
+      "step": 84
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.882591093117409e-06,
+      "loss": 1.5104,
+      "step": 85
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.963562753036438e-06,
+      "loss": 1.2957,
+      "step": 86
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.044534412955466e-06,
+      "loss": 1.4934,
+      "step": 87
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.125506072874494e-06,
+      "loss": 1.6621,
+      "step": 88
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.206477732793523e-06,
+      "loss": 1.4596,
+      "step": 89
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.2874493927125516e-06,
+      "loss": 1.5197,
+      "step": 90
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.368421052631579e-06,
+      "loss": 1.6523,
+      "step": 91
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.449392712550608e-06,
+      "loss": 1.4254,
+      "step": 92
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.5303643724696364e-06,
+      "loss": 1.5387,
+      "step": 93
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.611336032388664e-06,
+      "loss": 1.6028,
+      "step": 94
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.692307692307694e-06,
+      "loss": 1.3366,
+      "step": 95
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.773279352226721e-06,
+      "loss": 1.5634,
+      "step": 96
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.854251012145749e-06,
+      "loss": 1.3977,
+      "step": 97
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.935222672064778e-06,
+      "loss": 1.4515,
+      "step": 98
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.016194331983806e-06,
+      "loss": 1.4084,
+      "step": 99
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.097165991902834e-06,
+      "loss": 1.4108,
+      "step": 100
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.178137651821862e-06,
+      "loss": 1.624,
+      "step": 101
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.259109311740891e-06,
+      "loss": 1.3411,
+      "step": 102
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.340080971659919e-06,
+      "loss": 1.534,
+      "step": 103
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.421052631578948e-06,
+      "loss": 1.6633,
+      "step": 104
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.502024291497976e-06,
+      "loss": 1.3995,
+      "step": 105
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.582995951417005e-06,
+      "loss": 1.4912,
+      "step": 106
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.663967611336033e-06,
+      "loss": 1.4961,
+      "step": 107
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.744939271255063e-06,
+      "loss": 1.2731,
+      "step": 108
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.82591093117409e-06,
+      "loss": 1.3238,
+      "step": 109
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.906882591093118e-06,
+      "loss": 1.4453,
+      "step": 110
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.987854251012147e-06,
+      "loss": 1.5639,
+      "step": 111
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.068825910931175e-06,
+      "loss": 1.4275,
+      "step": 112
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.149797570850203e-06,
+      "loss": 1.3,
+      "step": 113
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.230769230769232e-06,
+      "loss": 1.7346,
+      "step": 114
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.31174089068826e-06,
+      "loss": 1.4611,
+      "step": 115
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.392712550607288e-06,
+      "loss": 1.386,
+      "step": 116
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.473684210526315e-06,
+      "loss": 1.4113,
+      "step": 117
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.554655870445345e-06,
+      "loss": 1.5984,
+      "step": 118
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.635627530364373e-06,
+      "loss": 1.4379,
+      "step": 119
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.7165991902834e-06,
+      "loss": 1.4222,
+      "step": 120
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.79757085020243e-06,
+      "loss": 1.5134,
+      "step": 121
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.878542510121458e-06,
+      "loss": 1.5347,
+      "step": 122
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.959514170040487e-06,
+      "loss": 1.6768,
+      "step": 123
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0040485829959515e-05,
+      "loss": 1.4213,
+      "step": 124
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0121457489878544e-05,
+      "loss": 1.5309,
+      "step": 125
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0202429149797572e-05,
+      "loss": 1.5515,
+      "step": 126
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.02834008097166e-05,
+      "loss": 1.5746,
+      "step": 127
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0364372469635629e-05,
+      "loss": 1.5103,
+      "step": 128
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0445344129554658e-05,
+      "loss": 1.4685,
+      "step": 129
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0526315789473684e-05,
+      "loss": 1.5195,
+      "step": 130
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0607287449392714e-05,
+      "loss": 1.7363,
+      "step": 131
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0688259109311743e-05,
+      "loss": 1.4953,
+      "step": 132
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.076923076923077e-05,
+      "loss": 1.3828,
+      "step": 133
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0850202429149799e-05,
+      "loss": 1.5473,
+      "step": 134
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0931174089068828e-05,
+      "loss": 1.4109,
+      "step": 135
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1012145748987854e-05,
+      "loss": 1.5874,
+      "step": 136
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1093117408906884e-05,
+      "loss": 1.4785,
+      "step": 137
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1174089068825913e-05,
+      "loss": 1.6936,
+      "step": 138
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1255060728744939e-05,
+      "loss": 1.4177,
+      "step": 139
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1336032388663969e-05,
+      "loss": 1.4616,
+      "step": 140
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1417004048582996e-05,
+      "loss": 1.4941,
+      "step": 141
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1497975708502026e-05,
+      "loss": 1.3895,
+      "step": 142
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1578947368421053e-05,
+      "loss": 1.3288,
+      "step": 143
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1659919028340081e-05,
+      "loss": 1.7203,
+      "step": 144
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.174089068825911e-05,
+      "loss": 1.5561,
+      "step": 145
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.182186234817814e-05,
+      "loss": 1.3653,
+      "step": 146
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1902834008097166e-05,
+      "loss": 1.5042,
+      "step": 147
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1983805668016195e-05,
+      "loss": 1.4304,
+      "step": 148
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2064777327935225e-05,
+      "loss": 1.3149,
+      "step": 149
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2145748987854251e-05,
+      "loss": 1.6766,
+      "step": 150
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.222672064777328e-05,
+      "loss": 1.3186,
+      "step": 151
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.230769230769231e-05,
+      "loss": 1.3491,
+      "step": 152
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2388663967611336e-05,
+      "loss": 1.4259,
+      "step": 153
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2469635627530365e-05,
+      "loss": 1.686,
+      "step": 154
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2550607287449395e-05,
+      "loss": 1.3956,
+      "step": 155
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.263157894736842e-05,
+      "loss": 1.5819,
+      "step": 156
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.271255060728745e-05,
+      "loss": 1.3796,
+      "step": 157
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.279352226720648e-05,
+      "loss": 1.5299,
+      "step": 158
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2874493927125507e-05,
+      "loss": 1.3695,
+      "step": 159
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2955465587044535e-05,
+      "loss": 1.331,
+      "step": 160
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3036437246963564e-05,
+      "loss": 1.4167,
+      "step": 161
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3117408906882592e-05,
+      "loss": 1.4637,
+      "step": 162
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3198380566801622e-05,
+      "loss": 1.4679,
+      "step": 163
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.327935222672065e-05,
+      "loss": 1.3731,
+      "step": 164
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3360323886639677e-05,
+      "loss": 1.5143,
+      "step": 165
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3441295546558706e-05,
+      "loss": 1.2914,
+      "step": 166
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3522267206477734e-05,
+      "loss": 1.1551,
+      "step": 167
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3603238866396762e-05,
+      "loss": 1.5316,
+      "step": 168
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3684210526315791e-05,
+      "loss": 1.5294,
+      "step": 169
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3765182186234817e-05,
+      "loss": 1.5303,
+      "step": 170
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3846153846153847e-05,
+      "loss": 1.3862,
+      "step": 171
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3927125506072876e-05,
+      "loss": 1.5622,
+      "step": 172
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4008097165991902e-05,
+      "loss": 1.4521,
+      "step": 173
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4089068825910932e-05,
+      "loss": 1.4824,
+      "step": 174
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4170040485829961e-05,
+      "loss": 1.4352,
+      "step": 175
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4251012145748989e-05,
+      "loss": 1.5346,
+      "step": 176
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4331983805668017e-05,
+      "loss": 1.3733,
+      "step": 177
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4412955465587046e-05,
+      "loss": 1.4553,
+      "step": 178
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4493927125506074e-05,
+      "loss": 1.4091,
+      "step": 179
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4574898785425103e-05,
+      "loss": 1.3432,
+      "step": 180
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.465587044534413e-05,
+      "loss": 1.6673,
+      "step": 181
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4736842105263159e-05,
+      "loss": 1.5151,
+      "step": 182
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4817813765182188e-05,
+      "loss": 1.4547,
+      "step": 183
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4898785425101216e-05,
+      "loss": 1.4961,
+      "step": 184
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4979757085020243e-05,
+      "loss": 1.4041,
+      "step": 185
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5060728744939273e-05,
+      "loss": 1.2743,
+      "step": 186
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5141700404858302e-05,
+      "loss": 1.2829,
+      "step": 187
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5222672064777328e-05,
+      "loss": 1.4366,
+      "step": 188
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5303643724696356e-05,
+      "loss": 1.4303,
+      "step": 189
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5384615384615387e-05,
+      "loss": 1.4877,
+      "step": 190
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5465587044534415e-05,
+      "loss": 1.5568,
+      "step": 191
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5546558704453443e-05,
+      "loss": 1.3585,
+      "step": 192
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.562753036437247e-05,
+      "loss": 1.2736,
+      "step": 193
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5708502024291498e-05,
+      "loss": 1.4784,
+      "step": 194
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.578947368421053e-05,
+      "loss": 1.5489,
+      "step": 195
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5870445344129557e-05,
+      "loss": 1.4311,
+      "step": 196
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5951417004048585e-05,
+      "loss": 1.3591,
+      "step": 197
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6032388663967612e-05,
+      "loss": 1.3625,
+      "step": 198
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6113360323886644e-05,
+      "loss": 1.4389,
+      "step": 199
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6194331983805668e-05,
+      "loss": 1.4999,
+      "step": 200
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.62753036437247e-05,
+      "loss": 1.4797,
+      "step": 201
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6356275303643723e-05,
+      "loss": 1.6856,
+      "step": 202
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6437246963562754e-05,
+      "loss": 1.3175,
+      "step": 203
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6518218623481782e-05,
+      "loss": 1.2925,
+      "step": 204
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.659919028340081e-05,
+      "loss": 1.4822,
+      "step": 205
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6680161943319838e-05,
+      "loss": 1.2931,
+      "step": 206
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.676113360323887e-05,
+      "loss": 1.6087,
+      "step": 207
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6842105263157896e-05,
+      "loss": 1.4546,
+      "step": 208
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6923076923076924e-05,
+      "loss": 1.3603,
+      "step": 209
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7004048582995952e-05,
+      "loss": 1.4072,
+      "step": 210
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.708502024291498e-05,
+      "loss": 1.5112,
+      "step": 211
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.716599190283401e-05,
+      "loss": 1.6165,
+      "step": 212
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.724696356275304e-05,
+      "loss": 1.4423,
+      "step": 213
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7327935222672066e-05,
+      "loss": 1.3194,
+      "step": 214
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7408906882591094e-05,
+      "loss": 1.6345,
+      "step": 215
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7489878542510125e-05,
+      "loss": 1.4448,
+      "step": 216
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.757085020242915e-05,
+      "loss": 1.3219,
+      "step": 217
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.765182186234818e-05,
+      "loss": 1.4405,
+      "step": 218
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7732793522267208e-05,
+      "loss": 1.3522,
+      "step": 219
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7813765182186236e-05,
+      "loss": 1.3755,
+      "step": 220
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7894736842105264e-05,
+      "loss": 1.5293,
+      "step": 221
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.7975708502024295e-05,
+      "loss": 1.3432,
+      "step": 222
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.805668016194332e-05,
+      "loss": 1.4495,
+      "step": 223
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.813765182186235e-05,
+      "loss": 1.4717,
+      "step": 224
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8218623481781378e-05,
+      "loss": 1.549,
+      "step": 225
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8299595141700406e-05,
+      "loss": 1.4257,
+      "step": 226
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8380566801619433e-05,
+      "loss": 1.4114,
+      "step": 227
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8461538461538465e-05,
+      "loss": 1.4764,
+      "step": 228
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8542510121457492e-05,
+      "loss": 1.3147,
+      "step": 229
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.862348178137652e-05,
+      "loss": 1.3793,
+      "step": 230
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8704453441295548e-05,
+      "loss": 1.6469,
+      "step": 231
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8785425101214576e-05,
+      "loss": 1.4886,
+      "step": 232
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8866396761133607e-05,
+      "loss": 1.4793,
+      "step": 233
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.894736842105263e-05,
+      "loss": 1.5703,
+      "step": 234
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9028340080971662e-05,
+      "loss": 1.3866,
+      "step": 235
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.910931174089069e-05,
+      "loss": 1.5349,
+      "step": 236
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9190283400809718e-05,
+      "loss": 1.2725,
+      "step": 237
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9271255060728745e-05,
+      "loss": 1.4084,
+      "step": 238
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9352226720647776e-05,
+      "loss": 1.6052,
+      "step": 239
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.94331983805668e-05,
+      "loss": 1.57,
+      "step": 240
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9514170040485832e-05,
+      "loss": 1.3511,
+      "step": 241
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.959514170040486e-05,
+      "loss": 1.4894,
+      "step": 242
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9676113360323887e-05,
+      "loss": 1.4141,
+      "step": 243
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9757085020242915e-05,
+      "loss": 1.2986,
+      "step": 244
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9838056680161946e-05,
+      "loss": 1.4149,
+      "step": 245
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9919028340080974e-05,
+      "loss": 1.7409,
+      "step": 246
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2e-05,
+      "loss": 1.424,
+      "step": 247
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999997745949496e-05,
+      "loss": 1.6062,
+      "step": 248
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999990983798994e-05,
+      "loss": 1.588,
+      "step": 249
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999979713551543e-05,
+      "loss": 1.4016,
+      "step": 250
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.999996393521222e-05,
+      "loss": 1.4154,
+      "step": 251
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999943648788148e-05,
+      "loss": 1.4393,
+      "step": 252
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.999991885428847e-05,
+      "loss": 1.412,
+      "step": 253
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999889551724358e-05,
+      "loss": 1.3357,
+      "step": 254
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999855741109022e-05,
+      "loss": 1.3643,
+      "step": 255
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999817422457707e-05,
+      "loss": 1.3624,
+      "step": 256
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999774595787688e-05,
+      "loss": 1.6102,
+      "step": 257
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999972726111827e-05,
+      "loss": 1.4927,
+      "step": 258
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999675418470796e-05,
+      "loss": 1.6233,
+      "step": 259
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999961906786863e-05,
+      "loss": 1.5101,
+      "step": 260
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999558209337182e-05,
+      "loss": 1.423,
+      "step": 261
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999492842903884e-05,
+      "loss": 1.5863,
+      "step": 262
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999422968598205e-05,
+      "loss": 1.4905,
+      "step": 263
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999348586451645e-05,
+      "loss": 1.4435,
+      "step": 264
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999269696497735e-05,
+      "loss": 1.5314,
+      "step": 265
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999186298772043e-05,
+      "loss": 1.2767,
+      "step": 266
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999909839331216e-05,
+      "loss": 1.6264,
+      "step": 267
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999900598015772e-05,
+      "loss": 1.4875,
+      "step": 268
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999890905935038e-05,
+      "loss": 1.3161,
+      "step": 269
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9998807630933834e-05,
+      "loss": 1.4557,
+      "step": 270
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999870169495381e-05,
+      "loss": 1.4756,
+      "step": 271
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999859125145806e-05,
+      "loss": 1.4169,
+      "step": 272
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9998476300496376e-05,
+      "loss": 1.4844,
+      "step": 273
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9998356842120577e-05,
+      "loss": 1.3807,
+      "step": 274
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999823287638452e-05,
+      "loss": 1.4012,
+      "step": 275
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9998104403344088e-05,
+      "loss": 1.6117,
+      "step": 276
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9997971423057193e-05,
+      "loss": 1.3178,
+      "step": 277
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999783393558379e-05,
+      "loss": 1.6779,
+      "step": 278
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999769194098586e-05,
+      "loss": 1.4029,
+      "step": 279
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9997545439327414e-05,
+      "loss": 1.8192,
+      "step": 280
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9997394430674495e-05,
+      "loss": 1.4888,
+      "step": 281
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9997238915095178e-05,
+      "loss": 1.5433,
+      "step": 282
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9997078892659578e-05,
+      "loss": 1.3345,
+      "step": 283
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9996914363439828e-05,
+      "loss": 1.4195,
+      "step": 284
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9996745327510098e-05,
+      "loss": 1.3985,
+      "step": 285
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9996571784946596e-05,
+      "loss": 1.2161,
+      "step": 286
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999639373582756e-05,
+      "loss": 1.6356,
+      "step": 287
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9996211180233246e-05,
+      "loss": 1.4707,
+      "step": 288
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999602411824596e-05,
+      "loss": 1.6066,
+      "step": 289
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999583254995003e-05,
+      "loss": 1.6023,
+      "step": 290
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999563647543181e-05,
+      "loss": 1.2656,
+      "step": 291
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9995435894779707e-05,
+      "loss": 1.3346,
+      "step": 292
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9995230808084133e-05,
+      "loss": 1.3497,
+      "step": 293
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9995021215437543e-05,
+      "loss": 1.5129,
+      "step": 294
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999480711693443e-05,
+      "loss": 1.5294,
+      "step": 295
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9994588512671303e-05,
+      "loss": 1.2934,
+      "step": 296
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9994365402746725e-05,
+      "loss": 1.3415,
+      "step": 297
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999413778726126e-05,
+      "loss": 1.6341,
+      "step": 298
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999390566631753e-05,
+      "loss": 1.4092,
+      "step": 299
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9993669040020176e-05,
+      "loss": 1.4249,
+      "step": 300
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999342790847587e-05,
+      "loss": 1.4611,
+      "step": 301
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9993182271793318e-05,
+      "loss": 1.4836,
+      "step": 302
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9992932130083248e-05,
+      "loss": 1.4917,
+      "step": 303
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9992677483458437e-05,
+      "loss": 1.3581,
+      "step": 304
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9992418332033677e-05,
+      "loss": 1.3556,
+      "step": 305
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9992154675925796e-05,
+      "loss": 1.4465,
+      "step": 306
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9991886515253654e-05,
+      "loss": 1.343,
+      "step": 307
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.999161385013814e-05,
+      "loss": 1.4941,
+      "step": 308
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9991336680702173e-05,
+      "loss": 1.3554,
+      "step": 309
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.999105500707071e-05,
+      "loss": 1.462,
+      "step": 310
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9990768829370722e-05,
+      "loss": 1.5492,
+      "step": 311
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.999047814773123e-05,
+      "loss": 1.2514,
+      "step": 312
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.999018296228327e-05,
+      "loss": 1.428,
+      "step": 313
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9989883273159916e-05,
+      "loss": 1.5197,
+      "step": 314
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9989579080496277e-05,
+      "loss": 1.4162,
+      "step": 315
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9989270384429474e-05,
+      "loss": 1.5225,
+      "step": 316
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.998895718509868e-05,
+      "loss": 1.3912,
+      "step": 317
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.998863948264509e-05,
+      "loss": 1.5477,
+      "step": 318
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.998831727721192e-05,
+      "loss": 1.4438,
+      "step": 319
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.998799056894443e-05,
+      "loss": 1.4114,
+      "step": 320
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9987659357989896e-05,
+      "loss": 1.6172,
+      "step": 321
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9987323644497636e-05,
+      "loss": 1.4603,
+      "step": 322
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9986983428618994e-05,
+      "loss": 1.5273,
+      "step": 323
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.998663871050734e-05,
+      "loss": 1.5221,
+      "step": 324
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9986289490318078e-05,
+      "loss": 1.4609,
+      "step": 325
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.998593576820864e-05,
+      "loss": 1.2795,
+      "step": 326
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9985577544338487e-05,
+      "loss": 1.661,
+      "step": 327
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9985214818869106e-05,
+      "loss": 1.4039,
+      "step": 328
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9984847591964026e-05,
+      "loss": 1.3859,
+      "step": 329
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.998447586378879e-05,
+      "loss": 1.5276,
+      "step": 330
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9984099634510983e-05,
+      "loss": 1.5926,
+      "step": 331
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9983718904300203e-05,
+      "loss": 1.3685,
+      "step": 332
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9983333673328095e-05,
+      "loss": 1.6115,
+      "step": 333
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9982943941768323e-05,
+      "loss": 1.2494,
+      "step": 334
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9982549709796582e-05,
+      "loss": 1.5279,
+      "step": 335
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9982150977590594e-05,
+      "loss": 1.3171,
+      "step": 336
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9981747745330112e-05,
+      "loss": 1.5477,
+      "step": 337
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.998134001319692e-05,
+      "loss": 1.4977,
+      "step": 338
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9980927781374827e-05,
+      "loss": 1.3219,
+      "step": 339
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.998051105004967e-05,
+      "loss": 1.4049,
+      "step": 340
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.998008981940931e-05,
+      "loss": 1.3808,
+      "step": 341
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9979664089643654e-05,
+      "loss": 1.4064,
+      "step": 342
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9979233860944617e-05,
+      "loss": 1.4659,
+      "step": 343
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9978799133506155e-05,
+      "loss": 1.486,
+      "step": 344
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9978359907524242e-05,
+      "loss": 1.3605,
+      "step": 345
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9977916183196887e-05,
+      "loss": 1.5471,
+      "step": 346
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.997746796072413e-05,
+      "loss": 1.5598,
+      "step": 347
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.997701524030803e-05,
+      "loss": 1.4577,
+      "step": 348
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9976558022152683e-05,
+      "loss": 1.5228,
+      "step": 349
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.99760963064642e-05,
+      "loss": 1.4021,
+      "step": 350
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.997563009345073e-05,
+      "loss": 1.2721,
+      "step": 351
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.997515938332245e-05,
+      "loss": 1.3869,
+      "step": 352
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9974684176291558e-05,
+      "loss": 1.3743,
+      "step": 353
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.997420447257228e-05,
+      "loss": 1.2245,
+      "step": 354
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9973720272380876e-05,
+      "loss": 1.3805,
+      "step": 355
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9973231575935623e-05,
+      "loss": 1.4318,
+      "step": 356
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9972738383456836e-05,
+      "loss": 1.5429,
+      "step": 357
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9972240695166846e-05,
+      "loss": 1.5375,
+      "step": 358
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9971738511290023e-05,
+      "loss": 1.2944,
+      "step": 359
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9971231832052747e-05,
+      "loss": 1.4951,
+      "step": 360
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.997072065768344e-05,
+      "loss": 1.3902,
+      "step": 361
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9970204988412546e-05,
+      "loss": 1.517,
+      "step": 362
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.996968482447253e-05,
+      "loss": 1.4421,
+      "step": 363
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9969160166097888e-05,
+      "loss": 1.4625,
+      "step": 364
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9968631013525143e-05,
+      "loss": 1.3002,
+      "step": 365
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9968097366992838e-05,
+      "loss": 1.374,
+      "step": 366
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9967559226741555e-05,
+      "loss": 1.3693,
+      "step": 367
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9967016593013883e-05,
+      "loss": 1.5158,
+      "step": 368
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9966469466054456e-05,
+      "loss": 1.3969,
+      "step": 369
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9965917846109913e-05,
+      "loss": 1.4014,
+      "step": 370
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9965361733428943e-05,
+      "loss": 1.4245,
+      "step": 371
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9964801128262237e-05,
+      "loss": 1.6052,
+      "step": 372
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9964236030862526e-05,
+      "loss": 1.1792,
+      "step": 373
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.996366644148456e-05,
+      "loss": 1.4428,
+      "step": 374
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.996309236038512e-05,
+      "loss": 1.3897,
+      "step": 375
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9962513787823003e-05,
+      "loss": 1.4706,
+      "step": 376
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9961930724059032e-05,
+      "loss": 1.7012,
+      "step": 377
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9961343169356068e-05,
+      "loss": 1.4478,
+      "step": 378
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9960751123978975e-05,
+      "loss": 1.56,
+      "step": 379
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9960154588194664e-05,
+      "loss": 1.4012,
+      "step": 380
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9959553562272057e-05,
+      "loss": 1.4373,
+      "step": 381
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9958948046482093e-05,
+      "loss": 1.386,
+      "step": 382
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9958338041097756e-05,
+      "loss": 1.4974,
+      "step": 383
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9957723546394036e-05,
+      "loss": 1.3908,
+      "step": 384
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9957104562647957e-05,
+      "loss": 1.4607,
+      "step": 385
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9956481090138562e-05,
+      "loss": 1.4329,
+      "step": 386
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9955853129146918e-05,
+      "loss": 1.3025,
+      "step": 387
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9955220679956118e-05,
+      "loss": 1.3846,
+      "step": 388
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9954583742851273e-05,
+      "loss": 1.4318,
+      "step": 389
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.995394231811952e-05,
+      "loss": 1.3613,
+      "step": 390
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9953296406050027e-05,
+      "loss": 1.507,
+      "step": 391
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9952646006933973e-05,
+      "loss": 1.3246,
+      "step": 392
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9951991121064566e-05,
+      "loss": 1.5635,
+      "step": 393
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.995133174873703e-05,
+      "loss": 1.4835,
+      "step": 394
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9950667890248624e-05,
+      "loss": 1.4813,
+      "step": 395
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9949999545898614e-05,
+      "loss": 1.6242,
+      "step": 396
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9949326715988304e-05,
+      "loss": 1.3476,
+      "step": 397
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9948649400821013e-05,
+      "loss": 1.438,
+      "step": 398
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9947967600702074e-05,
+      "loss": 1.4523,
+      "step": 399
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9947281315938854e-05,
+      "loss": 1.4365,
+      "step": 400
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9946590546840736e-05,
+      "loss": 1.5202,
+      "step": 401
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9945895293719125e-05,
+      "loss": 1.5144,
+      "step": 402
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9945195556887452e-05,
+      "loss": 1.6085,
+      "step": 403
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9944491336661164e-05,
+      "loss": 1.6904,
+      "step": 404
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.994378263335773e-05,
+      "loss": 1.374,
+      "step": 405
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9943069447296636e-05,
+      "loss": 1.5891,
+      "step": 406
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9942351778799402e-05,
+      "loss": 1.4125,
+      "step": 407
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9941629628189553e-05,
+      "loss": 1.5906,
+      "step": 408
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9940902995792644e-05,
+      "loss": 1.5653,
+      "step": 409
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9940171881936248e-05,
+      "loss": 1.3072,
+      "step": 410
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.993943628694996e-05,
+      "loss": 1.4744,
+      "step": 411
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9938696211165397e-05,
+      "loss": 1.2172,
+      "step": 412
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9937951654916182e-05,
+      "loss": 1.367,
+      "step": 413
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.993720261853798e-05,
+      "loss": 1.2985,
+      "step": 414
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9936449102368457e-05,
+      "loss": 1.5399,
+      "step": 415
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.993569110674731e-05,
+      "loss": 1.3786,
+      "step": 416
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9934928632016247e-05,
+      "loss": 1.4086,
+      "step": 417
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9934161678519e-05,
+      "loss": 1.7859,
+      "step": 418
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9933390246601324e-05,
+      "loss": 1.2968,
+      "step": 419
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9932614336610983e-05,
+      "loss": 1.4251,
+      "step": 420
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9931833948897768e-05,
+      "loss": 1.4614,
+      "step": 421
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9931049083813483e-05,
+      "loss": 1.5316,
+      "step": 422
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9930259741711958e-05,
+      "loss": 1.4388,
+      "step": 423
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.992946592294903e-05,
+      "loss": 1.5525,
+      "step": 424
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9928667627882567e-05,
+      "loss": 1.3703,
+      "step": 425
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9927864856872442e-05,
+      "loss": 1.533,
+      "step": 426
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9927057610280554e-05,
+      "loss": 1.4308,
+      "step": 427
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9926245888470824e-05,
+      "loss": 1.3886,
+      "step": 428
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9925429691809176e-05,
+      "loss": 1.29,
+      "step": 429
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9924609020663566e-05,
+      "loss": 1.2922,
+      "step": 430
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9923783875403957e-05,
+      "loss": 1.2831,
+      "step": 431
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9922954256402334e-05,
+      "loss": 1.2827,
+      "step": 432
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.99221201640327e-05,
+      "loss": 1.501,
+      "step": 433
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.992128159867107e-05,
+      "loss": 1.3091,
+      "step": 434
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9920438560695476e-05,
+      "loss": 1.3397,
+      "step": 435
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9919591050485966e-05,
+      "loss": 1.5943,
+      "step": 436
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.991873906842462e-05,
+      "loss": 1.421,
+      "step": 437
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.99178826148955e-05,
+      "loss": 1.4017,
+      "step": 438
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.991702169028472e-05,
+      "loss": 1.6338,
+      "step": 439
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9916156294980383e-05,
+      "loss": 1.2742,
+      "step": 440
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9915286429372626e-05,
+      "loss": 1.3077,
+      "step": 441
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.991441209385359e-05,
+      "loss": 1.3765,
+      "step": 442
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.991353328881743e-05,
+      "loss": 1.4766,
+      "step": 443
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9912650014660328e-05,
+      "loss": 1.2911,
+      "step": 444
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.991176227178047e-05,
+      "loss": 1.6949,
+      "step": 445
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9910870060578052e-05,
+      "loss": 1.53,
+      "step": 446
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9909973381455305e-05,
+      "loss": 1.4083,
+      "step": 447
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9909072234816455e-05,
+      "loss": 1.2591,
+      "step": 448
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9908166621067742e-05,
+      "loss": 1.5104,
+      "step": 449
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9907256540617435e-05,
+      "loss": 1.6514,
+      "step": 450
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.99063419938758e-05,
+      "loss": 1.3468,
+      "step": 451
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.990542298125513e-05,
+      "loss": 1.4805,
+      "step": 452
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9904499503169724e-05,
+      "loss": 1.516,
+      "step": 453
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9903571560035892e-05,
+      "loss": 1.6743,
+      "step": 454
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.990263915227196e-05,
+      "loss": 1.4625,
+      "step": 455
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.990170228029827e-05,
+      "loss": 1.2934,
+      "step": 456
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9900760944537178e-05,
+      "loss": 1.2859,
+      "step": 457
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9899815145413036e-05,
+      "loss": 1.3288,
+      "step": 458
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9898864883352228e-05,
+      "loss": 1.5507,
+      "step": 459
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9897910158783138e-05,
+      "loss": 1.4364,
+      "step": 460
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9896950972136165e-05,
+      "loss": 1.5114,
+      "step": 461
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.989598732384373e-05,
+      "loss": 1.2526,
+      "step": 462
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.989501921434024e-05,
+      "loss": 1.3621,
+      "step": 463
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9894046644062137e-05,
+      "loss": 1.4541,
+      "step": 464
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9893069613447867e-05,
+      "loss": 1.5295,
+      "step": 465
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.989208812293788e-05,
+      "loss": 1.1864,
+      "step": 466
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9891102172974647e-05,
+      "loss": 1.3131,
+      "step": 467
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.989011176400264e-05,
+      "loss": 1.3495,
+      "step": 468
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9889116896468348e-05,
+      "loss": 1.2759,
+      "step": 469
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9888117570820264e-05,
+      "loss": 1.512,
+      "step": 470
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.98871137875089e-05,
+      "loss": 1.5564,
+      "step": 471
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9886105546986765e-05,
+      "loss": 1.3214,
+      "step": 472
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.988509284970839e-05,
+      "loss": 1.7776,
+      "step": 473
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9884075696130304e-05,
+      "loss": 1.587,
+      "step": 474
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9883054086711054e-05,
+      "loss": 1.4746,
+      "step": 475
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.988202802191119e-05,
+      "loss": 1.3684,
+      "step": 476
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9880997502193275e-05,
+      "loss": 1.5468,
+      "step": 477
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9879962528021873e-05,
+      "loss": 1.5398,
+      "step": 478
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.987892309986356e-05,
+      "loss": 1.3812,
+      "step": 479
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.987787921818693e-05,
+      "loss": 1.8594,
+      "step": 480
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9876830883462565e-05,
+      "loss": 1.4741,
+      "step": 481
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.987577809616307e-05,
+      "loss": 1.3119,
+      "step": 482
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.987472085676305e-05,
+      "loss": 1.4137,
+      "step": 483
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9873659165739127e-05,
+      "loss": 1.4359,
+      "step": 484
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.987259302356991e-05,
+      "loss": 1.249,
+      "step": 485
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.987152243073603e-05,
+      "loss": 1.5577,
+      "step": 486
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.987044738772013e-05,
+      "loss": 1.3322,
+      "step": 487
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.986936789500684e-05,
+      "loss": 1.361,
+      "step": 488
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9868283953082808e-05,
+      "loss": 1.4352,
+      "step": 489
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9867195562436692e-05,
+      "loss": 1.1678,
+      "step": 490
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9866102723559143e-05,
+      "loss": 1.4805,
+      "step": 491
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9865005436942824e-05,
+      "loss": 1.337,
+      "step": 492
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9863903703082408e-05,
+      "loss": 1.3433,
+      "step": 493
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9862797522474565e-05,
+      "loss": 1.4536,
+      "step": 494
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.986168689561797e-05,
+      "loss": 1.4813,
+      "step": 495
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9860571823013307e-05,
+      "loss": 1.2977,
+      "step": 496
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9859452305163264e-05,
+      "loss": 1.287,
+      "step": 497
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9858328342572527e-05,
+      "loss": 1.4675,
+      "step": 498
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9857199935747794e-05,
+      "loss": 1.135,
+      "step": 499
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9856067085197756e-05,
+      "loss": 1.3397,
+      "step": 500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.985492979143312e-05,
+      "loss": 1.2142,
+      "step": 501
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9853788054966584e-05,
+      "loss": 1.4178,
+      "step": 502
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9852641876312863e-05,
+      "loss": 1.5688,
+      "step": 503
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9851491255988654e-05,
+      "loss": 1.4369,
+      "step": 504
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9850336194512674e-05,
+      "loss": 1.2237,
+      "step": 505
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.984917669240564e-05,
+      "loss": 1.3905,
+      "step": 506
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.984801275019026e-05,
+      "loss": 1.3532,
+      "step": 507
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.984684436839126e-05,
+      "loss": 1.5317,
+      "step": 508
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9845671547535347e-05,
+      "loss": 1.333,
+      "step": 509
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9844494288151254e-05,
+      "loss": 1.4554,
+      "step": 510
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9843312590769688e-05,
+      "loss": 1.4707,
+      "step": 511
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9842126455923377e-05,
+      "loss": 1.4948,
+      "step": 512
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9840935884147042e-05,
+      "loss": 1.5732,
+      "step": 513
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9839740875977407e-05,
+      "loss": 1.3753,
+      "step": 514
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9838541431953188e-05,
+      "loss": 1.404,
+      "step": 515
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.983733755261511e-05,
+      "loss": 1.5602,
+      "step": 516
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9836129238505898e-05,
+      "loss": 1.507,
+      "step": 517
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9834916490170265e-05,
+      "loss": 1.3977,
+      "step": 518
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9833699308154933e-05,
+      "loss": 1.4109,
+      "step": 519
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.983247769300862e-05,
+      "loss": 1.6683,
+      "step": 520
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9831251645282042e-05,
+      "loss": 1.381,
+      "step": 521
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9830021165527916e-05,
+      "loss": 1.3784,
+      "step": 522
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.982878625430095e-05,
+      "loss": 1.5702,
+      "step": 523
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9827546912157857e-05,
+      "loss": 1.3986,
+      "step": 524
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9826303139657347e-05,
+      "loss": 1.403,
+      "step": 525
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9825054937360126e-05,
+      "loss": 1.2385,
+      "step": 526
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.982380230582889e-05,
+      "loss": 1.4916,
+      "step": 527
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9822545245628345e-05,
+      "loss": 1.5525,
+      "step": 528
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9821283757325178e-05,
+      "loss": 1.4846,
+      "step": 529
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.982001784148809e-05,
+      "loss": 1.5945,
+      "step": 530
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9818747498687763e-05,
+      "loss": 1.4452,
+      "step": 531
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.981747272949688e-05,
+      "loss": 1.3749,
+      "step": 532
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9816193534490125e-05,
+      "loss": 1.4417,
+      "step": 533
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9814909914244168e-05,
+      "loss": 1.3764,
+      "step": 534
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9813621869337676e-05,
+      "loss": 1.3027,
+      "step": 535
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9812329400351318e-05,
+      "loss": 1.3641,
+      "step": 536
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9811032507867746e-05,
+      "loss": 1.3536,
+      "step": 537
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.980973119247162e-05,
+      "loss": 1.4129,
+      "step": 538
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.980842545474958e-05,
+      "loss": 1.3945,
+      "step": 539
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9807115295290268e-05,
+      "loss": 1.2565,
+      "step": 540
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9805800714684316e-05,
+      "loss": 1.2517,
+      "step": 541
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.980448171352435e-05,
+      "loss": 1.4449,
+      "step": 542
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9803158292404988e-05,
+      "loss": 1.4502,
+      "step": 543
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9801830451922844e-05,
+      "loss": 1.5581,
+      "step": 544
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.980049819267653e-05,
+      "loss": 1.4396,
+      "step": 545
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9799161515266623e-05,
+      "loss": 1.6378,
+      "step": 546
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9797820420295722e-05,
+      "loss": 1.5725,
+      "step": 547
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.979647490836841e-05,
+      "loss": 1.2291,
+      "step": 548
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.979512498009125e-05,
+      "loss": 1.5894,
+      "step": 549
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9793770636072803e-05,
+      "loss": 1.4543,
+      "step": 550
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9792411876923626e-05,
+      "loss": 1.4495,
+      "step": 551
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9791048703256262e-05,
+      "loss": 1.6203,
+      "step": 552
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.978968111568524e-05,
+      "loss": 1.4286,
+      "step": 553
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.978830911482708e-05,
+      "loss": 1.4958,
+      "step": 554
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9786932701300295e-05,
+      "loss": 1.5498,
+      "step": 555
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.978555187572539e-05,
+      "loss": 1.563,
+      "step": 556
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.978416663872485e-05,
+      "loss": 1.6036,
+      "step": 557
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9782776990923165e-05,
+      "loss": 1.4563,
+      "step": 558
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.978138293294679e-05,
+      "loss": 1.3453,
+      "step": 559
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.977998446542418e-05,
+      "loss": 1.2942,
+      "step": 560
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.977858158898579e-05,
+      "loss": 1.3698,
+      "step": 561
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9777174304264036e-05,
+      "loss": 1.4233,
+      "step": 562
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.977576261189335e-05,
+      "loss": 1.3717,
+      "step": 563
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9774346512510126e-05,
+      "loss": 1.2658,
+      "step": 564
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9772926006752763e-05,
+      "loss": 1.4578,
+      "step": 565
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.977150109526164e-05,
+      "loss": 1.3519,
+      "step": 566
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9770071778679116e-05,
+      "loss": 1.1714,
+      "step": 567
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9768638057649543e-05,
+      "loss": 1.5356,
+      "step": 568
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.976719993281926e-05,
+      "loss": 1.4522,
+      "step": 569
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9765757404836584e-05,
+      "loss": 1.4329,
+      "step": 570
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9764310474351824e-05,
+      "loss": 1.1701,
+      "step": 571
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9762859142017272e-05,
+      "loss": 1.4872,
+      "step": 572
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.97614034084872e-05,
+      "loss": 1.4089,
+      "step": 573
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9759943274417873e-05,
+      "loss": 1.2887,
+      "step": 574
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9758478740467523e-05,
+      "loss": 1.2367,
+      "step": 575
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9757009807296387e-05,
+      "loss": 1.2987,
+      "step": 576
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.975553647556667e-05,
+      "loss": 1.4055,
+      "step": 577
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.975405874594257e-05,
+      "loss": 1.4867,
+      "step": 578
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.975257661909026e-05,
+      "loss": 1.2507,
+      "step": 579
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.975109009567789e-05,
+      "loss": 1.5091,
+      "step": 580
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9749599176375613e-05,
+      "loss": 1.4957,
+      "step": 581
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9748103861855542e-05,
+      "loss": 1.223,
+      "step": 582
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.974660415279178e-05,
+      "loss": 1.4003,
+      "step": 583
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9745100049860417e-05,
+      "loss": 1.5182,
+      "step": 584
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.974359155373951e-05,
+      "loss": 1.3273,
+      "step": 585
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9742078665109107e-05,
+      "loss": 1.4104,
+      "step": 586
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9740561384651236e-05,
+      "loss": 1.3237,
+      "step": 587
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.97390397130499e-05,
+      "loss": 1.1254,
+      "step": 588
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9737513650991086e-05,
+      "loss": 1.4734,
+      "step": 589
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9735983199162758e-05,
+      "loss": 1.3547,
+      "step": 590
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9734448358254855e-05,
+      "loss": 1.2374,
+      "step": 591
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9732909128959306e-05,
+      "loss": 1.4385,
+      "step": 592
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9731365511970003e-05,
+      "loss": 1.4295,
+      "step": 593
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9729817507982828e-05,
+      "loss": 1.4355,
+      "step": 594
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9728265117695638e-05,
+      "loss": 1.5501,
+      "step": 595
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9726708341808268e-05,
+      "loss": 1.2357,
+      "step": 596
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.972514718102252e-05,
+      "loss": 1.2907,
+      "step": 597
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.972358163604219e-05,
+      "loss": 1.3888,
+      "step": 598
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9722011707573037e-05,
+      "loss": 1.5197,
+      "step": 599
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9720437396322802e-05,
+      "loss": 1.5389,
+      "step": 600
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.97188587030012e-05,
+      "loss": 1.4166,
+      "step": 601
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.971727562831992e-05,
+      "loss": 1.5481,
+      "step": 602
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.971568817299263e-05,
+      "loss": 1.3435,
+      "step": 603
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9714096337734974e-05,
+      "loss": 1.4975,
+      "step": 604
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9712500123264563e-05,
+      "loss": 1.4136,
+      "step": 605
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9710899530300987e-05,
+      "loss": 1.46,
+      "step": 606
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.970929455956581e-05,
+      "loss": 1.4882,
+      "step": 607
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9707685211782574e-05,
+      "loss": 1.3986,
+      "step": 608
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9706071487676778e-05,
+      "loss": 1.4502,
+      "step": 609
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9704453387975912e-05,
+      "loss": 1.4457,
+      "step": 610
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9702830913409437e-05,
+      "loss": 1.2928,
+      "step": 611
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.970120406470877e-05,
+      "loss": 1.2762,
+      "step": 612
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.969957284260732e-05,
+      "loss": 1.2404,
+      "step": 613
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9697937247840452e-05,
+      "loss": 1.4159,
+      "step": 614
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9696297281145514e-05,
+      "loss": 1.5292,
+      "step": 615
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9694652943261814e-05,
+      "loss": 1.3984,
+      "step": 616
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.969300423493064e-05,
+      "loss": 1.5631,
+      "step": 617
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9691351156895244e-05,
+      "loss": 1.5472,
+      "step": 618
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.968969370990085e-05,
+      "loss": 1.3806,
+      "step": 619
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9688031894694658e-05,
+      "loss": 1.4081,
+      "step": 620
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9686365712025822e-05,
+      "loss": 1.5174,
+      "step": 621
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9684695162645482e-05,
+      "loss": 1.5819,
+      "step": 622
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.968302024730673e-05,
+      "loss": 1.4613,
+      "step": 623
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9681340966764646e-05,
+      "loss": 1.6989,
+      "step": 624
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9679657321776256e-05,
+      "loss": 1.2583,
+      "step": 625
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.967796931310057e-05,
+      "loss": 1.6114,
+      "step": 626
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.967627694149856e-05,
+      "loss": 1.354,
+      "step": 627
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.967458020773316e-05,
+      "loss": 1.3605,
+      "step": 628
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9672879112569278e-05,
+      "loss": 1.4553,
+      "step": 629
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.967117365677378e-05,
+      "loss": 1.3442,
+      "step": 630
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.966946384111551e-05,
+      "loss": 1.6882,
+      "step": 631
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9667749666365267e-05,
+      "loss": 1.4265,
+      "step": 632
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9666031133295816e-05,
+      "loss": 1.2682,
+      "step": 633
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.966430824268189e-05,
+      "loss": 1.42,
+      "step": 634
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.966258099530019e-05,
+      "loss": 1.4029,
+      "step": 635
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.966084939192937e-05,
+      "loss": 1.2674,
+      "step": 636
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9659113433350057e-05,
+      "loss": 1.5612,
+      "step": 637
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.965737312034484e-05,
+      "loss": 1.3204,
+      "step": 638
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.965562845369827e-05,
+      "loss": 1.6025,
+      "step": 639
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9653879434196855e-05,
+      "loss": 1.449,
+      "step": 640
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9652126062629076e-05,
+      "loss": 1.4004,
+      "step": 641
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.965036833978537e-05,
+      "loss": 1.5046,
+      "step": 642
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.964860626645814e-05,
+      "loss": 1.6242,
+      "step": 643
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9646839843441734e-05,
+      "loss": 1.286,
+      "step": 644
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.964506907153249e-05,
+      "loss": 1.6114,
+      "step": 645
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9643293951528674e-05,
+      "loss": 1.4984,
+      "step": 646
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9641514484230538e-05,
+      "loss": 1.3505,
+      "step": 647
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.963973067044028e-05,
+      "loss": 1.7268,
+      "step": 648
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.963794251096206e-05,
+      "loss": 1.2968,
+      "step": 649
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9636150006602005e-05,
+      "loss": 1.4051,
+      "step": 650
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9634353158168188e-05,
+      "loss": 1.467,
+      "step": 651
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9632551966470646e-05,
+      "loss": 1.6361,
+      "step": 652
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.963074643232138e-05,
+      "loss": 1.3291,
+      "step": 653
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9628936556534337e-05,
+      "loss": 1.3833,
+      "step": 654
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9627122339925432e-05,
+      "loss": 1.4026,
+      "step": 655
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9625303783312527e-05,
+      "loss": 1.6362,
+      "step": 656
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.962348088751545e-05,
+      "loss": 1.1874,
+      "step": 657
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.962165365335598e-05,
+      "loss": 1.5393,
+      "step": 658
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.961982208165785e-05,
+      "loss": 1.3464,
+      "step": 659
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9617986173246758e-05,
+      "loss": 1.7629,
+      "step": 660
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.961614592895034e-05,
+      "loss": 1.4784,
+      "step": 661
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9614301349598204e-05,
+      "loss": 1.6267,
+      "step": 662
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9612452436021906e-05,
+      "loss": 1.2631,
+      "step": 663
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9610599189054947e-05,
+      "loss": 1.2433,
+      "step": 664
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9608741609532796e-05,
+      "loss": 1.4485,
+      "step": 665
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.960687969829287e-05,
+      "loss": 1.3422,
+      "step": 666
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9605013456174532e-05,
+      "loss": 1.4024,
+      "step": 667
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9603142884019105e-05,
+      "loss": 1.5671,
+      "step": 668
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.960126798266986e-05,
+      "loss": 1.5091,
+      "step": 669
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9599388752972027e-05,
+      "loss": 1.3844,
+      "step": 670
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9597505195772774e-05,
+      "loss": 1.4329,
+      "step": 671
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9595617311921237e-05,
+      "loss": 1.6001,
+      "step": 672
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.959372510226848e-05,
+      "loss": 1.3477,
+      "step": 673
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9591828567667547e-05,
+      "loss": 1.5072,
+      "step": 674
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.95899277089734e-05,
+      "loss": 1.3593,
+      "step": 675
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9588022527042975e-05,
+      "loss": 1.3783,
+      "step": 676
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.958611302273514e-05,
+      "loss": 1.483,
+      "step": 677
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9584199196910727e-05,
+      "loss": 1.3287,
+      "step": 678
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.95822810504325e-05,
+      "loss": 1.4167,
+      "step": 679
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.958035858416518e-05,
+      "loss": 1.2339,
+      "step": 680
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9578431798975442e-05,
+      "loss": 1.5815,
+      "step": 681
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9576500695731892e-05,
+      "loss": 1.3556,
+      "step": 682
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9574565275305096e-05,
+      "loss": 1.383,
+      "step": 683
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9572625538567558e-05,
+      "loss": 1.4768,
+      "step": 684
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.957068148639373e-05,
+      "loss": 1.646,
+      "step": 685
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.956873311966001e-05,
+      "loss": 1.2956,
+      "step": 686
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9566780439244746e-05,
+      "loss": 1.4599,
+      "step": 687
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9564823446028223e-05,
+      "loss": 1.2915,
+      "step": 688
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9562862140892673e-05,
+      "loss": 1.3599,
+      "step": 689
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9560896524722274e-05,
+      "loss": 1.5913,
+      "step": 690
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9558926598403144e-05,
+      "loss": 1.4244,
+      "step": 691
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9556952362823346e-05,
+      "loss": 1.34,
+      "step": 692
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9554973818872884e-05,
+      "loss": 1.6144,
+      "step": 693
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.955299096744371e-05,
+      "loss": 1.3717,
+      "step": 694
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9551003809429707e-05,
+      "loss": 1.4969,
+      "step": 695
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.954901234572671e-05,
+      "loss": 1.5831,
+      "step": 696
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9547016577232494e-05,
+      "loss": 1.5079,
+      "step": 697
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9545016504846765e-05,
+      "loss": 1.2387,
+      "step": 698
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.954301212947118e-05,
+      "loss": 1.3006,
+      "step": 699
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9541003452009327e-05,
+      "loss": 1.3181,
+      "step": 700
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9538990473366748e-05,
+      "loss": 1.4918,
+      "step": 701
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9536973194450903e-05,
+      "loss": 1.2795,
+      "step": 702
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9534951616171208e-05,
+      "loss": 1.5282,
+      "step": 703
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.953292573943901e-05,
+      "loss": 1.1994,
+      "step": 704
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9530895565167595e-05,
+      "loss": 1.5513,
+      "step": 705
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.952886109427218e-05,
+      "loss": 1.56,
+      "step": 706
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9526822327669933e-05,
+      "loss": 1.3745,
+      "step": 707
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.952477926627995e-05,
+      "loss": 1.3121,
+      "step": 708
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9522731911023258e-05,
+      "loss": 1.5165,
+      "step": 709
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.952068026282283e-05,
+      "loss": 1.4533,
+      "step": 710
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.951862432260357e-05,
+      "loss": 1.4485,
+      "step": 711
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9516564091292314e-05,
+      "loss": 1.4118,
+      "step": 712
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9514499569817835e-05,
+      "loss": 1.2783,
+      "step": 713
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9512430759110844e-05,
+      "loss": 1.3213,
+      "step": 714
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9510357660103975e-05,
+      "loss": 1.3319,
+      "step": 715
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.950828027373181e-05,
+      "loss": 1.3805,
+      "step": 716
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.950619860093085e-05,
+      "loss": 1.692,
+      "step": 717
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9504112642639538e-05,
+      "loss": 1.5939,
+      "step": 718
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.950202239979824e-05,
+      "loss": 1.3312,
+      "step": 719
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9499927873349268e-05,
+      "loss": 1.486,
+      "step": 720
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9497829064236842e-05,
+      "loss": 1.7123,
+      "step": 721
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9495725973407138e-05,
+      "loss": 1.5359,
+      "step": 722
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9493618601808244e-05,
+      "loss": 1.2175,
+      "step": 723
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.949150695039019e-05,
+      "loss": 1.5012,
+      "step": 724
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9489391020104922e-05,
+      "loss": 1.5308,
+      "step": 725
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.948727081190633e-05,
+      "loss": 1.2677,
+      "step": 726
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.948514632675022e-05,
+      "loss": 1.6378,
+      "step": 727
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9483017565594337e-05,
+      "loss": 1.5146,
+      "step": 728
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.948088452939834e-05,
+      "loss": 1.2962,
+      "step": 729
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9478747219123834e-05,
+      "loss": 1.483,
+      "step": 730
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9476605635734334e-05,
+      "loss": 1.521,
+      "step": 731
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9474459780195284e-05,
+      "loss": 1.4888,
+      "step": 732
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9472309653474065e-05,
+      "loss": 1.4327,
+      "step": 733
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.947015525653997e-05,
+      "loss": 1.3359,
+      "step": 734
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9467996590364224e-05,
+      "loss": 1.2588,
+      "step": 735
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9465833655919976e-05,
+      "loss": 1.2579,
+      "step": 736
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.94636664541823e-05,
+      "loss": 1.5636,
+      "step": 737
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.946149498612819e-05,
+      "loss": 1.4699,
+      "step": 738
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.945931925273657e-05,
+      "loss": 1.3783,
+      "step": 739
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9457139254988275e-05,
+      "loss": 1.5289,
+      "step": 740
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9454954993866078e-05,
+      "loss": 1.4068,
+      "step": 741
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9452766470354663e-05,
+      "loss": 1.3125,
+      "step": 742
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9450573685440636e-05,
+      "loss": 1.353,
+      "step": 743
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.944837664011253e-05,
+      "loss": 1.269,
+      "step": 744
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9446175335360796e-05,
+      "loss": 1.5075,
+      "step": 745
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.94439697721778e-05,
+      "loss": 1.6115,
+      "step": 746
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9441759951557838e-05,
+      "loss": 1.4908,
+      "step": 747
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9439545874497112e-05,
+      "loss": 1.2937,
+      "step": 748
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9437327541993755e-05,
+      "loss": 1.2214,
+      "step": 749
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9435104955047816e-05,
+      "loss": 1.2504,
+      "step": 750
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9432878114661252e-05,
+      "loss": 1.4884,
+      "step": 751
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9430647021837955e-05,
+      "loss": 1.6501,
+      "step": 752
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9428411677583717e-05,
+      "loss": 1.3908,
+      "step": 753
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9426172082906254e-05,
+      "loss": 1.2748,
+      "step": 754
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9423928238815202e-05,
+      "loss": 1.4199,
+      "step": 755
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9421680146322105e-05,
+      "loss": 1.4487,
+      "step": 756
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9419427806440428e-05,
+      "loss": 1.3828,
+      "step": 757
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9417171220185546e-05,
+      "loss": 1.587,
+      "step": 758
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9414910388574755e-05,
+      "loss": 1.1996,
+      "step": 759
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9412645312627258e-05,
+      "loss": 1.4598,
+      "step": 760
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9410375993364172e-05,
+      "loss": 1.5926,
+      "step": 761
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9408102431808533e-05,
+      "loss": 1.5198,
+      "step": 762
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9405824628985283e-05,
+      "loss": 1.2078,
+      "step": 763
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.940354258592128e-05,
+      "loss": 1.3316,
+      "step": 764
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9401256303645292e-05,
+      "loss": 1.4193,
+      "step": 765
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9398965783187997e-05,
+      "loss": 1.3933,
+      "step": 766
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9396671025581988e-05,
+      "loss": 1.4272,
+      "step": 767
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.939437203186176e-05,
+      "loss": 1.4964,
+      "step": 768
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9392068803063727e-05,
+      "loss": 1.3687,
+      "step": 769
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9389761340226205e-05,
+      "loss": 1.1247,
+      "step": 770
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.938744964438942e-05,
+      "loss": 1.498,
+      "step": 771
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9385133716595513e-05,
+      "loss": 1.418,
+      "step": 772
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9382813557888522e-05,
+      "loss": 1.2239,
+      "step": 773
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.93804891693144e-05,
+      "loss": 1.4466,
+      "step": 774
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.937816055192101e-05,
+      "loss": 1.4796,
+      "step": 775
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9375827706758107e-05,
+      "loss": 1.4391,
+      "step": 776
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9373490634877373e-05,
+      "loss": 1.4926,
+      "step": 777
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9371149337332368e-05,
+      "loss": 1.5284,
+      "step": 778
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9368803815178588e-05,
+      "loss": 1.2502,
+      "step": 779
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9366454069473408e-05,
+      "loss": 1.4235,
+      "step": 780
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.936410010127612e-05,
+      "loss": 1.339,
+      "step": 781
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.936174191164792e-05,
+      "loss": 1.3317,
+      "step": 782
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9359379501651897e-05,
+      "loss": 1.2711,
+      "step": 783
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9357012872353056e-05,
+      "loss": 1.5681,
+      "step": 784
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9354642024818293e-05,
+      "loss": 1.4495,
+      "step": 785
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9352266960116414e-05,
+      "loss": 1.5742,
+      "step": 786
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9349887679318117e-05,
+      "loss": 1.3013,
+      "step": 787
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.934750418349601e-05,
+      "loss": 1.5995,
+      "step": 788
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9345116473724595e-05,
+      "loss": 1.3671,
+      "step": 789
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9342724551080275e-05,
+      "loss": 1.3645,
+      "step": 790
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9340328416641353e-05,
+      "loss": 1.577,
+      "step": 791
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9337928071488036e-05,
+      "loss": 1.3929,
+      "step": 792
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.933552351670241e-05,
+      "loss": 1.414,
+      "step": 793
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9333114753368492e-05,
+      "loss": 1.4604,
+      "step": 794
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9330701782572165e-05,
+      "loss": 1.392,
+      "step": 795
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.932828460540122e-05,
+      "loss": 1.4222,
+      "step": 796
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.932586322294535e-05,
+      "loss": 1.3873,
+      "step": 797
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.932343763629613e-05,
+      "loss": 1.4244,
+      "step": 798
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.932100784654705e-05,
+      "loss": 1.5205,
+      "step": 799
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.931857385479348e-05,
+      "loss": 1.5153,
+      "step": 800
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9316135662132684e-05,
+      "loss": 1.2851,
+      "step": 801
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9313693269663827e-05,
+      "loss": 1.3228,
+      "step": 802
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9311246678487965e-05,
+      "loss": 1.5714,
+      "step": 803
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9308795889708046e-05,
+      "loss": 1.3657,
+      "step": 804
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9306340904428907e-05,
+      "loss": 1.4565,
+      "step": 805
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9303881723757285e-05,
+      "loss": 1.5162,
+      "step": 806
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.93014183488018e-05,
+      "loss": 1.3042,
+      "step": 807
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9298950780672963e-05,
+      "loss": 1.4244,
+      "step": 808
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9296479020483184e-05,
+      "loss": 1.4587,
+      "step": 809
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9294003069346764e-05,
+      "loss": 1.499,
+      "step": 810
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.929152292837987e-05,
+      "loss": 1.4951,
+      "step": 811
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.928903859870059e-05,
+      "loss": 1.4826,
+      "step": 812
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9286550081428875e-05,
+      "loss": 1.3317,
+      "step": 813
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9284057377686577e-05,
+      "loss": 1.4332,
+      "step": 814
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9281560488597436e-05,
+      "loss": 1.0955,
+      "step": 815
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.927905941528707e-05,
+      "loss": 1.2662,
+      "step": 816
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9276554158882986e-05,
+      "loss": 1.3177,
+      "step": 817
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9274044720514583e-05,
+      "loss": 1.4565,
+      "step": 818
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.927153110131314e-05,
+      "loss": 1.2897,
+      "step": 819
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9269013302411826e-05,
+      "loss": 1.3435,
+      "step": 820
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9266491324945686e-05,
+      "loss": 1.3053,
+      "step": 821
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.926396517005165e-05,
+      "loss": 1.4033,
+      "step": 822
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.926143483886854e-05,
+      "loss": 1.269,
+      "step": 823
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.925890033253705e-05,
+      "loss": 1.2307,
+      "step": 824
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9256361652199768e-05,
+      "loss": 1.3844,
+      "step": 825
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9253818799001146e-05,
+      "loss": 1.2856,
+      "step": 826
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9251271774087538e-05,
+      "loss": 1.2914,
+      "step": 827
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9248720578607165e-05,
+      "loss": 1.3426,
+      "step": 828
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.924616521371013e-05,
+      "loss": 1.5267,
+      "step": 829
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9243605680548422e-05,
+      "loss": 1.32,
+      "step": 830
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9241041980275895e-05,
+      "loss": 1.4527,
+      "step": 831
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.92384741140483e-05,
+      "loss": 1.5066,
+      "step": 832
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9235902083023254e-05,
+      "loss": 1.4001,
+      "step": 833
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9233325888360257e-05,
+      "loss": 1.3741,
+      "step": 834
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.923074553122068e-05,
+      "loss": 1.2338,
+      "step": 835
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.922816101276777e-05,
+      "loss": 1.4113,
+      "step": 836
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9225572334166666e-05,
+      "loss": 1.3984,
+      "step": 837
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9222979496584363e-05,
+      "loss": 1.5294,
+      "step": 838
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9220382501189738e-05,
+      "loss": 1.3275,
+      "step": 839
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9217781349153543e-05,
+      "loss": 1.4486,
+      "step": 840
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9215176041648403e-05,
+      "loss": 1.4433,
+      "step": 841
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9212566579848818e-05,
+      "loss": 1.1176,
+      "step": 842
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9209952964931164e-05,
+      "loss": 1.5473,
+      "step": 843
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.920733519807368e-05,
+      "loss": 1.2014,
+      "step": 844
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9204713280456483e-05,
+      "loss": 1.5584,
+      "step": 845
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.920208721326156e-05,
+      "loss": 1.4142,
+      "step": 846
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.919945699767277e-05,
+      "loss": 1.6049,
+      "step": 847
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.919682263487584e-05,
+      "loss": 1.3201,
+      "step": 848
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9194184126058363e-05,
+      "loss": 1.4475,
+      "step": 849
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9191541472409807e-05,
+      "loss": 1.4749,
+      "step": 850
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.918889467512151e-05,
+      "loss": 1.4961,
+      "step": 851
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9186243735386677e-05,
+      "loss": 1.3539,
+      "step": 852
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9183588654400377e-05,
+      "loss": 1.3759,
+      "step": 853
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.918092943335954e-05,
+      "loss": 1.3243,
+      "step": 854
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.917826607346298e-05,
+      "loss": 1.4738,
+      "step": 855
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.917559857591136e-05,
+      "loss": 1.627,
+      "step": 856
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.917292694190722e-05,
+      "loss": 1.4164,
+      "step": 857
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.917025117265495e-05,
+      "loss": 1.6502,
+      "step": 858
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9167571269360824e-05,
+      "loss": 1.363,
+      "step": 859
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9164887233232966e-05,
+      "loss": 1.4422,
+      "step": 860
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9162199065481368e-05,
+      "loss": 1.3528,
+      "step": 861
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.915950676731788e-05,
+      "loss": 1.3722,
+      "step": 862
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9156810339956213e-05,
+      "loss": 1.4398,
+      "step": 863
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9154109784611953e-05,
+      "loss": 1.5724,
+      "step": 864
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9151405102502538e-05,
+      "loss": 1.3364,
+      "step": 865
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9148696294847257e-05,
+      "loss": 1.5272,
+      "step": 866
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9145983362867278e-05,
+      "loss": 1.4079,
+      "step": 867
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9143266307785607e-05,
+      "loss": 1.3858,
+      "step": 868
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.914054513082713e-05,
+      "loss": 1.5023,
+      "step": 869
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9137819833218575e-05,
+      "loss": 1.2978,
+      "step": 870
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9135090416188536e-05,
+      "loss": 1.4008,
+      "step": 871
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9132356880967462e-05,
+      "loss": 1.2306,
+      "step": 872
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9129619228787657e-05,
+      "loss": 1.2276,
+      "step": 873
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9126877460883283e-05,
+      "loss": 1.2225,
+      "step": 874
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9124131578490357e-05,
+      "loss": 1.1401,
+      "step": 875
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.912138158284675e-05,
+      "loss": 1.248,
+      "step": 876
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.911862747519219e-05,
+      "loss": 1.2752,
+      "step": 877
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.911586925676825e-05,
+      "loss": 1.537,
+      "step": 878
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9113106928818367e-05,
+      "loss": 1.3029,
+      "step": 879
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9110340492587828e-05,
+      "loss": 1.3862,
+      "step": 880
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.910756994932377e-05,
+      "loss": 1.3654,
+      "step": 881
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9104795300275182e-05,
+      "loss": 1.4214,
+      "step": 882
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9102016546692904e-05,
+      "loss": 1.3243,
+      "step": 883
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.909923368982962e-05,
+      "loss": 1.3954,
+      "step": 884
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9096446730939877e-05,
+      "loss": 1.4737,
+      "step": 885
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9093655671280062e-05,
+      "loss": 1.614,
+      "step": 886
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9090860512108413e-05,
+      "loss": 1.2167,
+      "step": 887
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9088061254685017e-05,
+      "loss": 1.1405,
+      "step": 888
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9085257900271805e-05,
+      "loss": 1.5882,
+      "step": 889
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.908245045013256e-05,
+      "loss": 1.3433,
+      "step": 890
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9079638905532905e-05,
+      "loss": 1.2642,
+      "step": 891
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9076823267740318e-05,
+      "loss": 1.3177,
+      "step": 892
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9074003538024116e-05,
+      "loss": 1.3469,
+      "step": 893
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9071179717655455e-05,
+      "loss": 1.2783,
+      "step": 894
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.906835180790735e-05,
+      "loss": 1.5185,
+      "step": 895
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9065519810054648e-05,
+      "loss": 1.3465,
+      "step": 896
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9062683725374042e-05,
+      "loss": 1.3896,
+      "step": 897
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.905984355514407e-05,
+      "loss": 1.5916,
+      "step": 898
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9056999300645106e-05,
+      "loss": 1.6699,
+      "step": 899
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9054150963159366e-05,
+      "loss": 1.4947,
+      "step": 900
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.905129854397092e-05,
+      "loss": 1.3738,
+      "step": 901
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9048442044365658e-05,
+      "loss": 1.2656,
+      "step": 902
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9045581465631322e-05,
+      "loss": 1.6075,
+      "step": 903
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.904271680905749e-05,
+      "loss": 1.3492,
+      "step": 904
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9039848075935574e-05,
+      "loss": 1.4895,
+      "step": 905
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9036975267558835e-05,
+      "loss": 1.392,
+      "step": 906
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.903409838522236e-05,
+      "loss": 1.3634,
+      "step": 907
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9031217430223075e-05,
+      "loss": 1.3745,
+      "step": 908
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.902833240385975e-05,
+      "loss": 1.3977,
+      "step": 909
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9025443307432977e-05,
+      "loss": 1.5992,
+      "step": 910
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.902255014224519e-05,
+      "loss": 1.4732,
+      "step": 911
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9019652909600664e-05,
+      "loss": 1.3775,
+      "step": 912
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9016751610805494e-05,
+      "loss": 1.5096,
+      "step": 913
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.901384624716762e-05,
+      "loss": 1.3947,
+      "step": 914
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.90109368199968e-05,
+      "loss": 1.3239,
+      "step": 915
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9008023330604645e-05,
+      "loss": 1.282,
+      "step": 916
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9005105780304582e-05,
+      "loss": 1.2268,
+      "step": 917
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9002184170411864e-05,
+      "loss": 1.5166,
+      "step": 918
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8999258502243593e-05,
+      "loss": 1.3356,
+      "step": 919
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8996328777118685e-05,
+      "loss": 1.3934,
+      "step": 920
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8993394996357887e-05,
+      "loss": 1.3637,
+      "step": 921
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.899045716128378e-05,
+      "loss": 1.3239,
+      "step": 922
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8987515273220775e-05,
+      "loss": 1.4128,
+      "step": 923
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8984569333495096e-05,
+      "loss": 1.4618,
+      "step": 924
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8981619343434804e-05,
+      "loss": 1.4971,
+      "step": 925
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.897866530436979e-05,
+      "loss": 1.2826,
+      "step": 926
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8975707217631758e-05,
+      "loss": 1.2502,
+      "step": 927
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8972745084554246e-05,
+      "loss": 1.4955,
+      "step": 928
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8969778906472616e-05,
+      "loss": 1.3796,
+      "step": 929
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8966808684724048e-05,
+      "loss": 1.4555,
+      "step": 930
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.896383442064755e-05,
+      "loss": 1.6164,
+      "step": 931
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8960856115583947e-05,
+      "loss": 1.3332,
+      "step": 932
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8957873770875893e-05,
+      "loss": 1.1944,
+      "step": 933
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8954887387867854e-05,
+      "loss": 1.5217,
+      "step": 934
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8951896967906127e-05,
+      "loss": 1.3945,
+      "step": 935
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.894890251233882e-05,
+      "loss": 1.1783,
+      "step": 936
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.894590402251587e-05,
+      "loss": 1.4303,
+      "step": 937
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8942901499789016e-05,
+      "loss": 1.2855,
+      "step": 938
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8939894945511832e-05,
+      "loss": 1.4529,
+      "step": 939
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8936884361039704e-05,
+      "loss": 1.3315,
+      "step": 940
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8933869747729834e-05,
+      "loss": 1.4159,
+      "step": 941
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8930851106941235e-05,
+      "loss": 1.4523,
+      "step": 942
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8927828440034743e-05,
+      "loss": 1.3621,
+      "step": 943
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.892480174837301e-05,
+      "loss": 1.2238,
+      "step": 944
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.89217710333205e-05,
+      "loss": 1.3413,
+      "step": 945
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8918736296243482e-05,
+      "loss": 1.4338,
+      "step": 946
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8915697538510054e-05,
+      "loss": 1.3733,
+      "step": 947
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.891265476149011e-05,
+      "loss": 1.6046,
+      "step": 948
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.890960796655538e-05,
+      "loss": 1.6101,
+      "step": 949
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.890655715507937e-05,
+      "loss": 1.4401,
+      "step": 950
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8903502328437437e-05,
+      "loss": 1.3861,
+      "step": 951
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8900443488006713e-05,
+      "loss": 1.4916,
+      "step": 952
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.889738063516616e-05,
+      "loss": 1.4521,
+      "step": 953
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.889431377129654e-05,
+      "loss": 1.2038,
+      "step": 954
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8891242897780427e-05,
+      "loss": 1.4887,
+      "step": 955
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8888168016002203e-05,
+      "loss": 1.1133,
+      "step": 956
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.888508912734806e-05,
+      "loss": 1.596,
+      "step": 957
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.888200623320598e-05,
+      "loss": 1.489,
+      "step": 958
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8878919334965777e-05,
+      "loss": 1.2998,
+      "step": 959
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.887582843401905e-05,
+      "loss": 1.4629,
+      "step": 960
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8872733531759204e-05,
+      "loss": 1.6363,
+      "step": 961
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.886963462958146e-05,
+      "loss": 1.6112,
+      "step": 962
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8866531728882826e-05,
+      "loss": 1.3673,
+      "step": 963
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8863424831062126e-05,
+      "loss": 1.5647,
+      "step": 964
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.886031393751998e-05,
+      "loss": 1.3909,
+      "step": 965
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8857199049658814e-05,
+      "loss": 1.254,
+      "step": 966
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8854080168882847e-05,
+      "loss": 1.5123,
+      "step": 967
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8850957296598102e-05,
+      "loss": 1.2933,
+      "step": 968
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.88478304342124e-05,
+      "loss": 1.5136,
+      "step": 969
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8844699583135364e-05,
+      "loss": 1.1914,
+      "step": 970
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8841564744778416e-05,
+      "loss": 1.5145,
+      "step": 971
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8838425920554768e-05,
+      "loss": 1.6439,
+      "step": 972
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8835283111879435e-05,
+      "loss": 1.5382,
+      "step": 973
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.883213632016923e-05,
+      "loss": 1.2062,
+      "step": 974
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8828985546842754e-05,
+      "loss": 1.477,
+      "step": 975
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.882583079332041e-05,
+      "loss": 1.359,
+      "step": 976
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.882267206102439e-05,
+      "loss": 1.2623,
+      "step": 977
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.881950935137869e-05,
+      "loss": 1.7453,
+      "step": 978
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8816342665809082e-05,
+      "loss": 1.495,
+      "step": 979
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8813172005743142e-05,
+      "loss": 1.5265,
+      "step": 980
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8809997372610244e-05,
+      "loss": 1.3561,
+      "step": 981
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8806818767841533e-05,
+      "loss": 1.4261,
+      "step": 982
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8803636192869964e-05,
+      "loss": 1.2522,
+      "step": 983
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8800449649130273e-05,
+      "loss": 1.5731,
+      "step": 984
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.879725913805898e-05,
+      "loss": 1.5604,
+      "step": 985
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8794064661094404e-05,
+      "loss": 1.2072,
+      "step": 986
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.879086621967665e-05,
+      "loss": 1.3379,
+      "step": 987
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8787663815247604e-05,
+      "loss": 1.4043,
+      "step": 988
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.878445744925095e-05,
+      "loss": 1.6047,
+      "step": 989
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8781247123132137e-05,
+      "loss": 1.4493,
+      "step": 990
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.877803283833842e-05,
+      "loss": 1.5481,
+      "step": 991
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.877481459631883e-05,
+      "loss": 1.6449,
+      "step": 992
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8771592398524184e-05,
+      "loss": 1.2802,
+      "step": 993
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.876836624640708e-05,
+      "loss": 1.3267,
+      "step": 994
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.87651361414219e-05,
+      "loss": 1.4742,
+      "step": 995
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.876190208502481e-05,
+      "loss": 1.2897,
+      "step": 996
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8758664078673754e-05,
+      "loss": 1.3966,
+      "step": 997
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8755422123828455e-05,
+      "loss": 1.542,
+      "step": 998
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8752176221950426e-05,
+      "loss": 1.7059,
+      "step": 999
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8748926374502946e-05,
+      "loss": 1.5255,
+      "step": 1000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.874567258295108e-05,
+      "loss": 1.4791,
+      "step": 1001
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.874241484876167e-05,
+      "loss": 1.5053,
+      "step": 1002
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8739153173403338e-05,
+      "loss": 1.17,
+      "step": 1003
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.873588755834648e-05,
+      "loss": 1.5332,
+      "step": 1004
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8732618005063265e-05,
+      "loss": 1.584,
+      "step": 1005
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8729344515027644e-05,
+      "loss": 1.2823,
+      "step": 1006
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8726067089715338e-05,
+      "loss": 1.3238,
+      "step": 1007
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8722785730603846e-05,
+      "loss": 1.3161,
+      "step": 1008
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8719500439172434e-05,
+      "loss": 1.5625,
+      "step": 1009
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8716211216902145e-05,
+      "loss": 1.2338,
+      "step": 1010
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8712918065275795e-05,
+      "loss": 1.5012,
+      "step": 1011
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.870962098577797e-05,
+      "loss": 1.4165,
+      "step": 1012
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8706319979895028e-05,
+      "loss": 1.2942,
+      "step": 1013
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8703015049115088e-05,
+      "loss": 1.2755,
+      "step": 1014
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.869970619492806e-05,
+      "loss": 1.1752,
+      "step": 1015
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8696393418825595e-05,
+      "loss": 1.2145,
+      "step": 1016
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8693076722301135e-05,
+      "loss": 1.5891,
+      "step": 1017
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8689756106849877e-05,
+      "loss": 1.2479,
+      "step": 1018
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8686431573968793e-05,
+      "loss": 1.3455,
+      "step": 1019
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8683103125156605e-05,
+      "loss": 1.5791,
+      "step": 1020
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.867977076191382e-05,
+      "loss": 1.453,
+      "step": 1021
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.86764344857427e-05,
+      "loss": 1.3465,
+      "step": 1022
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.867309429814727e-05,
+      "loss": 1.3366,
+      "step": 1023
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.866975020063332e-05,
+      "loss": 1.4258,
+      "step": 1024
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8666402194708406e-05,
+      "loss": 1.4414,
+      "step": 1025
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.866305028188184e-05,
+      "loss": 1.7297,
+      "step": 1026
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.86596944636647e-05,
+      "loss": 1.2944,
+      "step": 1027
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8656334741569823e-05,
+      "loss": 1.5346,
+      "step": 1028
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8652971117111798e-05,
+      "loss": 1.4726,
+      "step": 1029
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8649603591806996e-05,
+      "loss": 1.3228,
+      "step": 1030
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8646232167173518e-05,
+      "loss": 1.5218,
+      "step": 1031
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.864285684473124e-05,
+      "loss": 1.4085,
+      "step": 1032
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8639477626001797e-05,
+      "loss": 1.3482,
+      "step": 1033
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8636094512508567e-05,
+      "loss": 1.3607,
+      "step": 1034
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8632707505776696e-05,
+      "loss": 1.3566,
+      "step": 1035
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8629316607333077e-05,
+      "loss": 1.4275,
+      "step": 1036
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.862592181870637e-05,
+      "loss": 1.3422,
+      "step": 1037
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.862252314142697e-05,
+      "loss": 1.4729,
+      "step": 1038
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8619120577027044e-05,
+      "loss": 1.5631,
+      "step": 1039
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8615714127040495e-05,
+      "loss": 1.4823,
+      "step": 1040
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.861230379300299e-05,
+      "loss": 1.2419,
+      "step": 1041
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8608889576451935e-05,
+      "loss": 1.5596,
+      "step": 1042
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8605471478926503e-05,
+      "loss": 1.5045,
+      "step": 1043
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.86020495019676e-05,
+      "loss": 1.3934,
+      "step": 1044
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.859862364711789e-05,
+      "loss": 1.329,
+      "step": 1045
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8595193915921786e-05,
+      "loss": 1.3949,
+      "step": 1046
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.859176030992544e-05,
+      "loss": 1.4682,
+      "step": 1047
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8588322830676758e-05,
+      "loss": 1.3905,
+      "step": 1048
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8584881479725387e-05,
+      "loss": 1.2528,
+      "step": 1049
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8581436258622733e-05,
+      "loss": 1.259,
+      "step": 1050
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8577987168921926e-05,
+      "loss": 1.4379,
+      "step": 1051
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8574534212177856e-05,
+      "loss": 1.3354,
+      "step": 1052
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8571077389947155e-05,
+      "loss": 1.3424,
+      "step": 1053
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8567616703788182e-05,
+      "loss": 1.5962,
+      "step": 1054
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8564152155261054e-05,
+      "loss": 1.4168,
+      "step": 1055
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.856068374592763e-05,
+      "loss": 1.3592,
+      "step": 1056
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8557211477351495e-05,
+      "loss": 1.3685,
+      "step": 1057
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.855373535109799e-05,
+      "loss": 1.4138,
+      "step": 1058
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8550255368734185e-05,
+      "loss": 1.4277,
+      "step": 1059
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.854677153182889e-05,
+      "loss": 1.4073,
+      "step": 1060
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8543283841952656e-05,
+      "loss": 1.4633,
+      "step": 1061
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.853979230067777e-05,
+      "loss": 1.1793,
+      "step": 1062
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.853629690957825e-05,
+      "loss": 1.344,
+      "step": 1063
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8532797670229858e-05,
+      "loss": 1.2717,
+      "step": 1064
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.852929458421008e-05,
+      "loss": 1.348,
+      "step": 1065
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.852578765309815e-05,
+      "loss": 1.4544,
+      "step": 1066
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8522276878475022e-05,
+      "loss": 1.3704,
+      "step": 1067
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8518762261923396e-05,
+      "loss": 1.3231,
+      "step": 1068
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.851524380502769e-05,
+      "loss": 1.401,
+      "step": 1069
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.851172150937406e-05,
+      "loss": 1.4641,
+      "step": 1070
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8508195376550394e-05,
+      "loss": 1.489,
+      "step": 1071
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.850466540814631e-05,
+      "loss": 1.3344,
+      "step": 1072
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8501131605753152e-05,
+      "loss": 1.2065,
+      "step": 1073
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8497593970963996e-05,
+      "loss": 1.4063,
+      "step": 1074
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8494052505373638e-05,
+      "loss": 1.4241,
+      "step": 1075
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8490507210578612e-05,
+      "loss": 1.4888,
+      "step": 1076
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.848695808817717e-05,
+      "loss": 1.44,
+      "step": 1077
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8483405139769293e-05,
+      "loss": 1.3683,
+      "step": 1078
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8479848366956686e-05,
+      "loss": 1.4267,
+      "step": 1079
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8476287771342777e-05,
+      "loss": 1.7991,
+      "step": 1080
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8472723354532718e-05,
+      "loss": 1.4608,
+      "step": 1081
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8469155118133388e-05,
+      "loss": 1.2808,
+      "step": 1082
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.846558306375338e-05,
+      "loss": 1.5795,
+      "step": 1083
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8462007193003016e-05,
+      "loss": 1.4844,
+      "step": 1084
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8458427507494332e-05,
+      "loss": 1.5345,
+      "step": 1085
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8454844008841088e-05,
+      "loss": 1.3195,
+      "step": 1086
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8451256698658755e-05,
+      "loss": 1.2367,
+      "step": 1087
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8447665578564534e-05,
+      "loss": 1.498,
+      "step": 1088
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8444070650177343e-05,
+      "loss": 1.4186,
+      "step": 1089
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8440471915117806e-05,
+      "loss": 1.5095,
+      "step": 1090
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8436869375008267e-05,
+      "loss": 1.4912,
+      "step": 1091
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8433263031472794e-05,
+      "loss": 1.4142,
+      "step": 1092
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8429652886137158e-05,
+      "loss": 1.3118,
+      "step": 1093
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.842603894062885e-05,
+      "loss": 1.4374,
+      "step": 1094
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8422421196577074e-05,
+      "loss": 1.4775,
+      "step": 1095
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.841879965561275e-05,
+      "loss": 1.0879,
+      "step": 1096
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8415174319368496e-05,
+      "loss": 1.3002,
+      "step": 1097
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8411545189478656e-05,
+      "loss": 1.5263,
+      "step": 1098
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.840791226757928e-05,
+      "loss": 1.3396,
+      "step": 1099
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8404275555308116e-05,
+      "loss": 1.3309,
+      "step": 1100
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8400635054304645e-05,
+      "loss": 1.5088,
+      "step": 1101
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.839699076621003e-05,
+      "loss": 1.473,
+      "step": 1102
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8393342692667163e-05,
+      "loss": 1.3119,
+      "step": 1103
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.838969083532062e-05,
+      "loss": 1.4572,
+      "step": 1104
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8386035195816706e-05,
+      "loss": 1.4027,
+      "step": 1105
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8382375775803413e-05,
+      "loss": 1.4873,
+      "step": 1106
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.837871257693045e-05,
+      "loss": 1.366,
+      "step": 1107
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.837504560084922e-05,
+      "loss": 1.4704,
+      "step": 1108
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.837137484921284e-05,
+      "loss": 1.2,
+      "step": 1109
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.836770032367611e-05,
+      "loss": 1.5934,
+      "step": 1110
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.836402202589555e-05,
+      "loss": 1.3195,
+      "step": 1111
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8360339957529375e-05,
+      "loss": 1.3545,
+      "step": 1112
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.83566541202375e-05,
+      "loss": 1.5754,
+      "step": 1113
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8352964515681532e-05,
+      "loss": 1.3804,
+      "step": 1114
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8349271145524785e-05,
+      "loss": 1.45,
+      "step": 1115
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8345574011432266e-05,
+      "loss": 1.5587,
+      "step": 1116
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8341873115070684e-05,
+      "loss": 1.6367,
+      "step": 1117
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8338168458108438e-05,
+      "loss": 1.4973,
+      "step": 1118
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8334460042215623e-05,
+      "loss": 1.514,
+      "step": 1119
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8330747869064036e-05,
+      "loss": 1.3391,
+      "step": 1120
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8327031940327156e-05,
+      "loss": 1.1679,
+      "step": 1121
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8323312257680163e-05,
+      "loss": 1.4111,
+      "step": 1122
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.831958882279993e-05,
+      "loss": 1.4273,
+      "step": 1123
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8315861637365017e-05,
+      "loss": 1.2995,
+      "step": 1124
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8312130703055675e-05,
+      "loss": 1.3981,
+      "step": 1125
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8308396021553853e-05,
+      "loss": 1.2627,
+      "step": 1126
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8304657594543173e-05,
+      "loss": 1.4253,
+      "step": 1127
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8300915423708968e-05,
+      "loss": 1.6773,
+      "step": 1128
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8297169510738238e-05,
+      "loss": 1.3796,
+      "step": 1129
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.829341985731968e-05,
+      "loss": 1.3386,
+      "step": 1130
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8289666465143673e-05,
+      "loss": 1.3795,
+      "step": 1131
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8285909335902294e-05,
+      "loss": 1.1607,
+      "step": 1132
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8282148471289285e-05,
+      "loss": 1.487,
+      "step": 1133
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8278383873000085e-05,
+      "loss": 1.3164,
+      "step": 1134
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.827461554273181e-05,
+      "loss": 1.358,
+      "step": 1135
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.827084348218327e-05,
+      "loss": 1.6349,
+      "step": 1136
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8267067693054937e-05,
+      "loss": 1.4547,
+      "step": 1137
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8263288177048984e-05,
+      "loss": 1.3851,
+      "step": 1138
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.825950493586925e-05,
+      "loss": 1.4479,
+      "step": 1139
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8255717971221257e-05,
+      "loss": 1.4941,
+      "step": 1140
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.825192728481221e-05,
+      "loss": 1.26,
+      "step": 1141
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.824813287835099e-05,
+      "loss": 1.4447,
+      "step": 1142
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.824433475354815e-05,
+      "loss": 1.5215,
+      "step": 1143
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8240532912115927e-05,
+      "loss": 1.4232,
+      "step": 1144
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8236727355768225e-05,
+      "loss": 1.4542,
+      "step": 1145
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.823291808622063e-05,
+      "loss": 1.3151,
+      "step": 1146
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8229105105190404e-05,
+      "loss": 1.3486,
+      "step": 1147
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8225288414396467e-05,
+      "loss": 1.3752,
+      "step": 1148
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8221468015559427e-05,
+      "loss": 1.3915,
+      "step": 1149
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.821764391040156e-05,
+      "loss": 1.5325,
+      "step": 1150
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8213816100646808e-05,
+      "loss": 1.2533,
+      "step": 1151
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.820998458802079e-05,
+      "loss": 1.5101,
+      "step": 1152
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8206149374250787e-05,
+      "loss": 1.4157,
+      "step": 1153
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8202310461065753e-05,
+      "loss": 1.2294,
+      "step": 1154
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.819846785019631e-05,
+      "loss": 1.3114,
+      "step": 1155
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8194621543374744e-05,
+      "loss": 1.426,
+      "step": 1156
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.819077154233501e-05,
+      "loss": 1.5947,
+      "step": 1157
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.818691784881273e-05,
+      "loss": 1.3127,
+      "step": 1158
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8183060464545184e-05,
+      "loss": 1.5112,
+      "step": 1159
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.817919939127132e-05,
+      "loss": 1.2744,
+      "step": 1160
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8175334630731753e-05,
+      "loss": 1.2704,
+      "step": 1161
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8171466184668755e-05,
+      "loss": 1.5041,
+      "step": 1162
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8167594054826253e-05,
+      "loss": 1.3295,
+      "step": 1163
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8163718242949852e-05,
+      "loss": 1.254,
+      "step": 1164
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.81598387507868e-05,
+      "loss": 1.3782,
+      "step": 1165
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.815595558008602e-05,
+      "loss": 1.5909,
+      "step": 1166
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8152068732598072e-05,
+      "loss": 1.5958,
+      "step": 1167
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8148178210075196e-05,
+      "loss": 1.4419,
+      "step": 1168
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8144284014271277e-05,
+      "loss": 1.2054,
+      "step": 1169
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8140386146941854e-05,
+      "loss": 1.2605,
+      "step": 1170
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8136484609844133e-05,
+      "loss": 1.5351,
+      "step": 1171
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8132579404736956e-05,
+      "loss": 1.4092,
+      "step": 1172
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8128670533380837e-05,
+      "loss": 1.4739,
+      "step": 1173
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.812475799753793e-05,
+      "loss": 1.4455,
+      "step": 1174
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8120841798972047e-05,
+      "loss": 1.3802,
+      "step": 1175
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.811692193944865e-05,
+      "loss": 1.2372,
+      "step": 1176
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8112998420734853e-05,
+      "loss": 1.4317,
+      "step": 1177
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8109071244599415e-05,
+      "loss": 1.5681,
+      "step": 1178
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.810514041281275e-05,
+      "loss": 1.33,
+      "step": 1179
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.810120592714691e-05,
+      "loss": 1.3569,
+      "step": 1180
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.809726778937561e-05,
+      "loss": 1.5121,
+      "step": 1181
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8093326001274198e-05,
+      "loss": 1.2911,
+      "step": 1182
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8089380564619673e-05,
+      "loss": 1.389,
+      "step": 1183
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8085431481190675e-05,
+      "loss": 1.5073,
+      "step": 1184
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.808147875276749e-05,
+      "loss": 1.5313,
+      "step": 1185
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.807752238113205e-05,
+      "loss": 1.2217,
+      "step": 1186
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.807356236806793e-05,
+      "loss": 1.275,
+      "step": 1187
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8069598715360342e-05,
+      "loss": 1.6806,
+      "step": 1188
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8065631424796137e-05,
+      "loss": 1.3558,
+      "step": 1189
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.806166049816381e-05,
+      "loss": 1.3332,
+      "step": 1190
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8057685937253503e-05,
+      "loss": 1.5453,
+      "step": 1191
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.805370774385698e-05,
+      "loss": 1.2452,
+      "step": 1192
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8049725919767652e-05,
+      "loss": 1.4377,
+      "step": 1193
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8045740466780574e-05,
+      "loss": 1.3176,
+      "step": 1194
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8041751386692414e-05,
+      "loss": 1.4339,
+      "step": 1195
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8037758681301502e-05,
+      "loss": 1.3947,
+      "step": 1196
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.803376235240778e-05,
+      "loss": 1.3593,
+      "step": 1197
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.802976240181284e-05,
+      "loss": 1.5346,
+      "step": 1198
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8025758831319898e-05,
+      "loss": 1.5941,
+      "step": 1199
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8021751642733805e-05,
+      "loss": 1.651,
+      "step": 1200
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8017740837861046e-05,
+      "loss": 1.3714,
+      "step": 1201
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8013726418509724e-05,
+      "loss": 1.5011,
+      "step": 1202
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.800970838648958e-05,
+      "loss": 1.4583,
+      "step": 1203
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.800568674361199e-05,
+      "loss": 1.1909,
+      "step": 1204
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.800166149168995e-05,
+      "loss": 1.4494,
+      "step": 1205
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7997632632538082e-05,
+      "loss": 1.3738,
+      "step": 1206
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7993600167972637e-05,
+      "loss": 1.3871,
+      "step": 1207
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7989564099811492e-05,
+      "loss": 1.3889,
+      "step": 1208
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7985524429874145e-05,
+      "loss": 1.5069,
+      "step": 1209
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.798148115998172e-05,
+      "loss": 1.4883,
+      "step": 1210
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7977434291956964e-05,
+      "loss": 1.4718,
+      "step": 1211
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7973383827624248e-05,
+      "loss": 1.3996,
+      "step": 1212
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7969329768809563e-05,
+      "loss": 1.432,
+      "step": 1213
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7965272117340517e-05,
+      "loss": 1.2937,
+      "step": 1214
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.796121087504634e-05,
+      "loss": 1.5314,
+      "step": 1215
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7957146043757885e-05,
+      "loss": 1.3218,
+      "step": 1216
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7953077625307612e-05,
+      "loss": 1.3523,
+      "step": 1217
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7949005621529613e-05,
+      "loss": 1.2405,
+      "step": 1218
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7944930034259583e-05,
+      "loss": 1.17,
+      "step": 1219
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.794085086533484e-05,
+      "loss": 1.5047,
+      "step": 1220
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7936768116594315e-05,
+      "loss": 1.5222,
+      "step": 1221
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.793268178987855e-05,
+      "loss": 1.4606,
+      "step": 1222
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7928591887029706e-05,
+      "loss": 1.4674,
+      "step": 1223
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7924498409891548e-05,
+      "loss": 1.3156,
+      "step": 1224
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.792040136030946e-05,
+      "loss": 1.4179,
+      "step": 1225
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7916300740130433e-05,
+      "loss": 1.3238,
+      "step": 1226
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.791219655120307e-05,
+      "loss": 1.3583,
+      "step": 1227
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7908088795377577e-05,
+      "loss": 1.5106,
+      "step": 1228
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.790397747450577e-05,
+      "loss": 1.3044,
+      "step": 1229
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.789986259044108e-05,
+      "loss": 1.2748,
+      "step": 1230
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7895744145038535e-05,
+      "loss": 1.4186,
+      "step": 1231
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7891622140154774e-05,
+      "loss": 1.3204,
+      "step": 1232
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7887496577648037e-05,
+      "loss": 1.4062,
+      "step": 1233
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7883367459378167e-05,
+      "loss": 1.3003,
+      "step": 1234
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7879234787206613e-05,
+      "loss": 1.1556,
+      "step": 1235
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.787509856299643e-05,
+      "loss": 1.3398,
+      "step": 1236
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7870958788612266e-05,
+      "loss": 1.3547,
+      "step": 1237
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7866815465920374e-05,
+      "loss": 1.364,
+      "step": 1238
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7862668596788603e-05,
+      "loss": 1.4831,
+      "step": 1239
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7858518183086406e-05,
+      "loss": 1.4051,
+      "step": 1240
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.785436422668483e-05,
+      "loss": 1.7197,
+      "step": 1241
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7850206729456524e-05,
+      "loss": 1.5809,
+      "step": 1242
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7846045693275727e-05,
+      "loss": 1.4733,
+      "step": 1243
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7841881120018274e-05,
+      "loss": 1.4481,
+      "step": 1244
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.78377130115616e-05,
+      "loss": 1.3704,
+      "step": 1245
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7833541369784732e-05,
+      "loss": 1.3213,
+      "step": 1246
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7829366196568283e-05,
+      "loss": 1.382,
+      "step": 1247
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7825187493794466e-05,
+      "loss": 1.4501,
+      "step": 1248
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7821005263347085e-05,
+      "loss": 1.3877,
+      "step": 1249
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7816819507111528e-05,
+      "loss": 1.5116,
+      "step": 1250
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7812630226974775e-05,
+      "loss": 1.4124,
+      "step": 1251
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7808437424825405e-05,
+      "loss": 1.457,
+      "step": 1252
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7804241102553562e-05,
+      "loss": 1.3536,
+      "step": 1253
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7800041262051e-05,
+      "loss": 1.4142,
+      "step": 1254
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7795837905211044e-05,
+      "loss": 1.3916,
+      "step": 1255
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7791631033928615e-05,
+      "loss": 1.2225,
+      "step": 1256
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.778742065010021e-05,
+      "loss": 1.317,
+      "step": 1257
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7783206755623915e-05,
+      "loss": 1.3769,
+      "step": 1258
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.777898935239939e-05,
+      "loss": 1.2045,
+      "step": 1259
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7774768442327893e-05,
+      "loss": 1.5615,
+      "step": 1260
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.777054402731224e-05,
+      "loss": 1.3093,
+      "step": 1261
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7766316109256852e-05,
+      "loss": 1.4622,
+      "step": 1262
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7762084690067713e-05,
+      "loss": 1.2823,
+      "step": 1263
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7757849771652387e-05,
+      "loss": 1.3706,
+      "step": 1264
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.775361135592002e-05,
+      "loss": 1.3934,
+      "step": 1265
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7749369444781332e-05,
+      "loss": 1.2523,
+      "step": 1266
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.774512404014862e-05,
+      "loss": 1.44,
+      "step": 1267
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7740875143935756e-05,
+      "loss": 1.5347,
+      "step": 1268
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7736622758058183e-05,
+      "loss": 1.4579,
+      "step": 1269
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.773236688443292e-05,
+      "loss": 1.3791,
+      "step": 1270
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7728107524978562e-05,
+      "loss": 1.4547,
+      "step": 1271
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7723844681615268e-05,
+      "loss": 1.1943,
+      "step": 1272
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.771957835626477e-05,
+      "loss": 1.4171,
+      "step": 1273
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.771530855085037e-05,
+      "loss": 1.3869,
+      "step": 1274
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7711035267296943e-05,
+      "loss": 1.3065,
+      "step": 1275
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7706758507530924e-05,
+      "loss": 1.4639,
+      "step": 1276
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7702478273480323e-05,
+      "loss": 1.223,
+      "step": 1277
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.769819456707471e-05,
+      "loss": 1.3004,
+      "step": 1278
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7693907390245225e-05,
+      "loss": 1.3684,
+      "step": 1279
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.768961674492457e-05,
+      "loss": 1.3791,
+      "step": 1280
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.768532263304702e-05,
+      "loss": 1.5456,
+      "step": 1281
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7681025056548384e-05,
+      "loss": 1.3621,
+      "step": 1282
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7676724017366068e-05,
+      "loss": 1.0241,
+      "step": 1283
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.767241951743902e-05,
+      "loss": 1.4697,
+      "step": 1284
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7668111558707755e-05,
+      "loss": 1.4707,
+      "step": 1285
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.766380014311434e-05,
+      "loss": 1.4878,
+      "step": 1286
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7659485272602407e-05,
+      "loss": 1.5917,
+      "step": 1287
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7655166949117138e-05,
+      "loss": 1.3739,
+      "step": 1288
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7650845174605288e-05,
+      "loss": 1.1107,
+      "step": 1289
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7646519951015145e-05,
+      "loss": 1.3867,
+      "step": 1290
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7642191280296576e-05,
+      "loss": 1.5891,
+      "step": 1291
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.763785916440097e-05,
+      "loss": 1.4058,
+      "step": 1292
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.763352360528131e-05,
+      "loss": 1.5218,
+      "step": 1293
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.76291846048921e-05,
+      "loss": 1.3751,
+      "step": 1294
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7624842165189405e-05,
+      "loss": 1.2756,
+      "step": 1295
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7620496288130838e-05,
+      "loss": 1.3693,
+      "step": 1296
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.761614697567557e-05,
+      "loss": 1.5033,
+      "step": 1297
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.761179422978431e-05,
+      "loss": 1.3553,
+      "step": 1298
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7607438052419326e-05,
+      "loss": 1.4684,
+      "step": 1299
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.760307844554442e-05,
+      "loss": 1.5294,
+      "step": 1300
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7598715411124953e-05,
+      "loss": 1.296,
+      "step": 1301
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.759434895112782e-05,
+      "loss": 1.4076,
+      "step": 1302
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.758997906752147e-05,
+      "loss": 1.4046,
+      "step": 1303
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7585605762275886e-05,
+      "loss": 1.3889,
+      "step": 1304
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.75812290373626e-05,
+      "loss": 1.389,
+      "step": 1305
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7576848894754684e-05,
+      "loss": 1.263,
+      "step": 1306
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.757246533642675e-05,
+      "loss": 1.5753,
+      "step": 1307
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.756807836435495e-05,
+      "loss": 1.3877,
+      "step": 1308
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7563687980516978e-05,
+      "loss": 1.1921,
+      "step": 1309
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7559294186892063e-05,
+      "loss": 1.6471,
+      "step": 1310
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7554896985460966e-05,
+      "loss": 1.3854,
+      "step": 1311
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7550496378205998e-05,
+      "loss": 1.276,
+      "step": 1312
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.754609236711099e-05,
+      "loss": 1.6011,
+      "step": 1313
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.754168495416132e-05,
+      "loss": 1.2046,
+      "step": 1314
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.753727414134389e-05,
+      "loss": 1.5555,
+      "step": 1315
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.753285993064714e-05,
+      "loss": 1.2937,
+      "step": 1316
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7528442324061044e-05,
+      "loss": 1.5105,
+      "step": 1317
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7524021323577104e-05,
+      "loss": 1.3756,
+      "step": 1318
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7519596931188344e-05,
+      "loss": 1.4909,
+      "step": 1319
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7515169148889334e-05,
+      "loss": 1.262,
+      "step": 1320
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7510737978676154e-05,
+      "loss": 1.436,
+      "step": 1321
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.750630342254643e-05,
+      "loss": 1.4236,
+      "step": 1322
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7501865482499298e-05,
+      "loss": 1.3528,
+      "step": 1323
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.749742416053543e-05,
+      "loss": 1.4658,
+      "step": 1324
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7492979458657014e-05,
+      "loss": 1.5084,
+      "step": 1325
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7488531378867773e-05,
+      "loss": 1.2585,
+      "step": 1326
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7484079923172942e-05,
+      "loss": 1.3755,
+      "step": 1327
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7479625093579278e-05,
+      "loss": 1.4061,
+      "step": 1328
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.747516689209507e-05,
+      "loss": 1.1276,
+      "step": 1329
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7470705320730122e-05,
+      "loss": 1.4733,
+      "step": 1330
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.746624038149575e-05,
+      "loss": 1.3383,
+      "step": 1331
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7461772076404794e-05,
+      "loss": 1.3923,
+      "step": 1332
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7457300407471616e-05,
+      "loss": 1.5212,
+      "step": 1333
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7452825376712084e-05,
+      "loss": 1.4116,
+      "step": 1334
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.744834698614359e-05,
+      "loss": 1.3437,
+      "step": 1335
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7443865237785037e-05,
+      "loss": 1.4438,
+      "step": 1336
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.743938013365684e-05,
+      "loss": 1.4078,
+      "step": 1337
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.743489167578093e-05,
+      "loss": 1.3974,
+      "step": 1338
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7430399866180756e-05,
+      "loss": 1.2118,
+      "step": 1339
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7425904706881263e-05,
+      "loss": 1.4608,
+      "step": 1340
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7421406199908915e-05,
+      "loss": 1.4252,
+      "step": 1341
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7416904347291694e-05,
+      "loss": 1.4052,
+      "step": 1342
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7412399151059066e-05,
+      "loss": 1.159,
+      "step": 1343
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.740789061324203e-05,
+      "loss": 1.2502,
+      "step": 1344
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7403378735873073e-05,
+      "loss": 1.5694,
+      "step": 1345
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.73988635209862e-05,
+      "loss": 1.4735,
+      "step": 1346
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7394344970616915e-05,
+      "loss": 1.3979,
+      "step": 1347
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7389823086802226e-05,
+      "loss": 1.3157,
+      "step": 1348
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.738529787158064e-05,
+      "loss": 1.557,
+      "step": 1349
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.738076932699217e-05,
+      "loss": 1.2214,
+      "step": 1350
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7376237455078337e-05,
+      "loss": 1.3252,
+      "step": 1351
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.737170225788215e-05,
+      "loss": 1.3674,
+      "step": 1352
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7367163737448118e-05,
+      "loss": 1.3697,
+      "step": 1353
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7362621895822258e-05,
+      "loss": 1.2826,
+      "step": 1354
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7358076735052075e-05,
+      "loss": 1.2991,
+      "step": 1355
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7353528257186575e-05,
+      "loss": 1.2942,
+      "step": 1356
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7348976464276255e-05,
+      "loss": 1.2115,
+      "step": 1357
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7344421358373108e-05,
+      "loss": 1.4956,
+      "step": 1358
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7339862941530632e-05,
+      "loss": 1.2678,
+      "step": 1359
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7335301215803794e-05,
+      "loss": 1.4708,
+      "step": 1360
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7330736183249075e-05,
+      "loss": 1.6024,
+      "step": 1361
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7326167845924432e-05,
+      "loss": 1.5789,
+      "step": 1362
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7321596205889324e-05,
+      "loss": 1.4116,
+      "step": 1363
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.731702126520469e-05,
+      "loss": 1.1826,
+      "step": 1364
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7312443025932955e-05,
+      "loss": 1.4338,
+      "step": 1365
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7307861490138038e-05,
+      "loss": 1.464,
+      "step": 1366
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7303276659885346e-05,
+      "loss": 1.1797,
+      "step": 1367
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.729868853724176e-05,
+      "loss": 1.4449,
+      "step": 1368
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.729409712427566e-05,
+      "loss": 1.5068,
+      "step": 1369
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.728950242305689e-05,
+      "loss": 1.3901,
+      "step": 1370
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7284904435656802e-05,
+      "loss": 1.4736,
+      "step": 1371
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.72803031641482e-05,
+      "loss": 1.372,
+      "step": 1372
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.7275698610605397e-05,
+      "loss": 1.1323,
+      "step": 1373
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.7271090777104168e-05,
+      "loss": 1.4681,
+      "step": 1374
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.7266479665721764e-05,
+      "loss": 1.5074,
+      "step": 1375
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.7261865278536928e-05,
+      "loss": 1.2574,
+      "step": 1376
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.7257247617629872e-05,
+      "loss": 1.1672,
+      "step": 1377
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.7252626685082283e-05,
+      "loss": 1.5069,
+      "step": 1378
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.724800248297732e-05,
+      "loss": 1.2797,
+      "step": 1379
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.7243375013399627e-05,
+      "loss": 1.3479,
+      "step": 1380
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.723874427843531e-05,
+      "loss": 1.3618,
+      "step": 1381
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.723411028017195e-05,
+      "loss": 1.5687,
+      "step": 1382
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.7229473020698603e-05,
+      "loss": 1.4223,
+      "step": 1383
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.7224832502105792e-05,
+      "loss": 1.2853,
+      "step": 1384
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.7220188726485507e-05,
+      "loss": 1.3332,
+      "step": 1385
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.7215541695931214e-05,
+      "loss": 1.5018,
+      "step": 1386
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.7210891412537836e-05,
+      "loss": 1.2428,
+      "step": 1387
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.720623787840177e-05,
+      "loss": 1.3516,
+      "step": 1388
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.7201581095620878e-05,
+      "loss": 1.6064,
+      "step": 1389
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.7196921066294482e-05,
+      "loss": 1.5437,
+      "step": 1390
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.7192257792523372e-05,
+      "loss": 1.5952,
+      "step": 1391
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.7187591276409795e-05,
+      "loss": 1.4435,
+      "step": 1392
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.7182921520057472e-05,
+      "loss": 1.426,
+      "step": 1393
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.717824852557157e-05,
+      "loss": 1.329,
+      "step": 1394
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.717357229505872e-05,
+      "loss": 1.5667,
+      "step": 1395
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.7168892830627016e-05,
+      "loss": 1.3297,
+      "step": 1396
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.7164210134386014e-05,
+      "loss": 1.1727,
+      "step": 1397
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.7159524208446714e-05,
+      "loss": 1.3517,
+      "step": 1398
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.715483505492158e-05,
+      "loss": 1.3949,
+      "step": 1399
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.715014267592453e-05,
+      "loss": 1.3825,
+      "step": 1400
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.714544707357094e-05,
+      "loss": 1.3756,
+      "step": 1401
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.714074824997763e-05,
+      "loss": 1.5185,
+      "step": 1402
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.7136046207262873e-05,
+      "loss": 1.5205,
+      "step": 1403
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.7131340947546406e-05,
+      "loss": 1.2443,
+      "step": 1404
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.71266324729494e-05,
+      "loss": 1.4229,
+      "step": 1405
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.7121920785594494e-05,
+      "loss": 1.5147,
+      "step": 1406
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.7117205887605757e-05,
+      "loss": 1.2557,
+      "step": 1407
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.711248778110871e-05,
+      "loss": 1.3483,
+      "step": 1408
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.7107766468230326e-05,
+      "loss": 1.5591,
+      "step": 1409
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.7103041951099023e-05,
+      "loss": 1.1117,
+      "step": 1410
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.7098314231844658e-05,
+      "loss": 1.2109,
+      "step": 1411
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.7093583312598534e-05,
+      "loss": 1.5371,
+      "step": 1412
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.7088849195493397e-05,
+      "loss": 1.3421,
+      "step": 1413
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.708411188266344e-05,
+      "loss": 1.1939,
+      "step": 1414
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.707937137624429e-05,
+      "loss": 1.3084,
+      "step": 1415
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.7074627678373005e-05,
+      "loss": 1.4072,
+      "step": 1416
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.70698807911881e-05,
+      "loss": 1.3419,
+      "step": 1417
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.706513071682953e-05,
+      "loss": 1.6337,
+      "step": 1418
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.7060377457438656e-05,
+      "loss": 1.352,
+      "step": 1419
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.705562101515831e-05,
+      "loss": 1.6078,
+      "step": 1420
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7050861392132737e-05,
+      "loss": 1.4048,
+      "step": 1421
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7046098590507627e-05,
+      "loss": 1.323,
+      "step": 1422
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.70413326124301e-05,
+      "loss": 1.3451,
+      "step": 1423
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7036563460048704e-05,
+      "loss": 1.4254,
+      "step": 1424
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7031791135513417e-05,
+      "loss": 1.4602,
+      "step": 1425
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7027015640975665e-05,
+      "loss": 1.2934,
+      "step": 1426
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7022236978588273e-05,
+      "loss": 1.3128,
+      "step": 1427
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7017455150505518e-05,
+      "loss": 1.4013,
+      "step": 1428
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.70126701588831e-05,
+      "loss": 1.3394,
+      "step": 1429
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.700788200587814e-05,
+      "loss": 1.2911,
+      "step": 1430
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7003090693649174e-05,
+      "loss": 1.4767,
+      "step": 1431
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.699829622435619e-05,
+      "loss": 1.3244,
+      "step": 1432
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6993498600160574e-05,
+      "loss": 1.3223,
+      "step": 1433
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6988697823225146e-05,
+      "loss": 1.2025,
+      "step": 1434
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6983893895714146e-05,
+      "loss": 1.5734,
+      "step": 1435
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6979086819793228e-05,
+      "loss": 1.2228,
+      "step": 1436
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6974276597629475e-05,
+      "loss": 1.4006,
+      "step": 1437
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6969463231391386e-05,
+      "loss": 1.1742,
+      "step": 1438
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.696464672324887e-05,
+      "loss": 1.549,
+      "step": 1439
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6959827075373258e-05,
+      "loss": 1.4887,
+      "step": 1440
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.69550042899373e-05,
+      "loss": 1.3278,
+      "step": 1441
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6950178369115155e-05,
+      "loss": 1.3854,
+      "step": 1442
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.694534931508239e-05,
+      "loss": 1.4669,
+      "step": 1443
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6940517130016e-05,
+      "loss": 1.4752,
+      "step": 1444
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6935681816094377e-05,
+      "loss": 1.2445,
+      "step": 1445
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6930843375497337e-05,
+      "loss": 1.5006,
+      "step": 1446
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.692600181040609e-05,
+      "loss": 1.5004,
+      "step": 1447
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6921157123003267e-05,
+      "loss": 1.2449,
+      "step": 1448
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6916309315472902e-05,
+      "loss": 1.2486,
+      "step": 1449
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.691145839000043e-05,
+      "loss": 1.4355,
+      "step": 1450
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.69066043487727e-05,
+      "loss": 1.5493,
+      "step": 1451
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6901747193977968e-05,
+      "loss": 1.4846,
+      "step": 1452
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6896886927805883e-05,
+      "loss": 1.3186,
+      "step": 1453
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6892023552447505e-05,
+      "loss": 1.3587,
+      "step": 1454
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6887157070095292e-05,
+      "loss": 1.4515,
+      "step": 1455
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6882287482943103e-05,
+      "loss": 1.3796,
+      "step": 1456
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6877414793186192e-05,
+      "loss": 1.2702,
+      "step": 1457
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6872539003021226e-05,
+      "loss": 1.481,
+      "step": 1458
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6867660114646257e-05,
+      "loss": 1.3841,
+      "step": 1459
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6862778130260732e-05,
+      "loss": 1.4066,
+      "step": 1460
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6857893052065507e-05,
+      "loss": 1.2261,
+      "step": 1461
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6853004882262817e-05,
+      "loss": 1.3011,
+      "step": 1462
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6848113623056302e-05,
+      "loss": 1.4456,
+      "step": 1463
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6843219276650996e-05,
+      "loss": 1.5532,
+      "step": 1464
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6838321845253313e-05,
+      "loss": 1.4879,
+      "step": 1465
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6833421331071065e-05,
+      "loss": 1.3918,
+      "step": 1466
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6828517736313455e-05,
+      "loss": 1.2437,
+      "step": 1467
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6823611063191073e-05,
+      "loss": 1.4728,
+      "step": 1468
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6818701313915897e-05,
+      "loss": 1.1929,
+      "step": 1469
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.681378849070129e-05,
+      "loss": 1.5301,
+      "step": 1470
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6808872595762003e-05,
+      "loss": 1.4303,
+      "step": 1471
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6803953631314173e-05,
+      "loss": 1.2321,
+      "step": 1472
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.679903159957532e-05,
+      "loss": 1.3044,
+      "step": 1473
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6794106502764336e-05,
+      "loss": 1.3406,
+      "step": 1474
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.678917834310152e-05,
+      "loss": 1.0745,
+      "step": 1475
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6784247122808518e-05,
+      "loss": 1.2977,
+      "step": 1476
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.677931284410839e-05,
+      "loss": 1.6242,
+      "step": 1477
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.677437550922555e-05,
+      "loss": 1.5949,
+      "step": 1478
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6769435120385803e-05,
+      "loss": 1.2384,
+      "step": 1479
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.676449167981632e-05,
+      "loss": 1.4403,
+      "step": 1480
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6759545189745664e-05,
+      "loss": 1.3228,
+      "step": 1481
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.675459565240375e-05,
+      "loss": 1.4841,
+      "step": 1482
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.674964307002189e-05,
+      "loss": 1.3202,
+      "step": 1483
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6744687444832753e-05,
+      "loss": 1.4606,
+      "step": 1484
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6739728779070386e-05,
+      "loss": 1.6489,
+      "step": 1485
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6734767074970203e-05,
+      "loss": 1.5467,
+      "step": 1486
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6729802334768995e-05,
+      "loss": 1.3538,
+      "step": 1487
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6724834560704917e-05,
+      "loss": 1.3005,
+      "step": 1488
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.671986375501749e-05,
+      "loss": 1.562,
+      "step": 1489
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6714889919947598e-05,
+      "loss": 1.1749,
+      "step": 1490
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6709913057737502e-05,
+      "loss": 1.5314,
+      "step": 1491
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6704933170630827e-05,
+      "loss": 1.3408,
+      "step": 1492
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6699950260872544e-05,
+      "loss": 1.5144,
+      "step": 1493
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.669496433070901e-05,
+      "loss": 1.361,
+      "step": 1494
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6689975382387924e-05,
+      "loss": 1.3733,
+      "step": 1495
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.668498341815836e-05,
+      "loss": 1.3058,
+      "step": 1496
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6679988440270745e-05,
+      "loss": 1.2407,
+      "step": 1497
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6674990450976863e-05,
+      "loss": 1.5351,
+      "step": 1498
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6669989452529864e-05,
+      "loss": 1.4354,
+      "step": 1499
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6664985447184237e-05,
+      "loss": 1.5059,
+      "step": 1500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6659978437195852e-05,
+      "loss": 1.4197,
+      "step": 1501
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.665496842482191e-05,
+      "loss": 1.2712,
+      "step": 1502
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.664995541232098e-05,
+      "loss": 1.2993,
+      "step": 1503
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.664493940195298e-05,
+      "loss": 1.6097,
+      "step": 1504
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.663992039597917e-05,
+      "loss": 1.1353,
+      "step": 1505
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6634898396662175e-05,
+      "loss": 1.3805,
+      "step": 1506
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6629873406265964e-05,
+      "loss": 1.4397,
+      "step": 1507
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.662484542705585e-05,
+      "loss": 1.5416,
+      "step": 1508
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6619814461298498e-05,
+      "loss": 1.3221,
+      "step": 1509
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.661478051126192e-05,
+      "loss": 1.2421,
+      "step": 1510
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6609743579215467e-05,
+      "loss": 1.4042,
+      "step": 1511
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.660470366742984e-05,
+      "loss": 1.1142,
+      "step": 1512
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.659966077817709e-05,
+      "loss": 1.2224,
+      "step": 1513
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6594614913730595e-05,
+      "loss": 1.4655,
+      "step": 1514
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6589566076365077e-05,
+      "loss": 1.3791,
+      "step": 1515
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6584514268356614e-05,
+      "loss": 1.4219,
+      "step": 1516
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6579459491982608e-05,
+      "loss": 1.446,
+      "step": 1517
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6574401749521797e-05,
+      "loss": 1.4257,
+      "step": 1518
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6569341043254272e-05,
+      "loss": 1.376,
+      "step": 1519
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6564277375461445e-05,
+      "loss": 1.5325,
+      "step": 1520
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6559210748426067e-05,
+      "loss": 1.4315,
+      "step": 1521
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6554141164432233e-05,
+      "loss": 1.3292,
+      "step": 1522
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6549068625765352e-05,
+      "loss": 1.4082,
+      "step": 1523
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.654399313471218e-05,
+      "loss": 1.4699,
+      "step": 1524
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6538914693560803e-05,
+      "loss": 1.2756,
+      "step": 1525
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.653383330460063e-05,
+      "loss": 1.2965,
+      "step": 1526
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6528748970122404e-05,
+      "loss": 1.3678,
+      "step": 1527
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.652366169241819e-05,
+      "loss": 1.5713,
+      "step": 1528
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6518571473781393e-05,
+      "loss": 1.2644,
+      "step": 1529
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.651347831650673e-05,
+      "loss": 1.3789,
+      "step": 1530
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.650838222289024e-05,
+      "loss": 1.6894,
+      "step": 1531
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.650328319522931e-05,
+      "loss": 1.5921,
+      "step": 1532
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6498181235822618e-05,
+      "loss": 1.4684,
+      "step": 1533
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6493076346970185e-05,
+      "loss": 1.2025,
+      "step": 1534
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.648796853097335e-05,
+      "loss": 1.3121,
+      "step": 1535
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6482857790134763e-05,
+      "loss": 1.3968,
+      "step": 1536
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.64777441267584e-05,
+      "loss": 1.3301,
+      "step": 1537
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.647262754314955e-05,
+      "loss": 1.2044,
+      "step": 1538
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.646750804161482e-05,
+      "loss": 1.4473,
+      "step": 1539
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6462385624462137e-05,
+      "loss": 1.226,
+      "step": 1540
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6457260294000738e-05,
+      "loss": 1.3208,
+      "step": 1541
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.645213205254117e-05,
+      "loss": 1.2196,
+      "step": 1542
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.64470009023953e-05,
+      "loss": 1.5405,
+      "step": 1543
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6441866845876296e-05,
+      "loss": 1.4634,
+      "step": 1544
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6436729885298647e-05,
+      "loss": 1.526,
+      "step": 1545
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6431590022978154e-05,
+      "loss": 1.3723,
+      "step": 1546
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6426447261231904e-05,
+      "loss": 1.5613,
+      "step": 1547
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6421301602378313e-05,
+      "loss": 1.5107,
+      "step": 1548
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.64161530487371e-05,
+      "loss": 1.245,
+      "step": 1549
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.641100160262928e-05,
+      "loss": 1.6679,
+      "step": 1550
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6405847266377178e-05,
+      "loss": 1.3472,
+      "step": 1551
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6400690042304423e-05,
+      "loss": 1.3675,
+      "step": 1552
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.639552993273594e-05,
+      "loss": 1.4041,
+      "step": 1553
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.639036693999796e-05,
+      "loss": 1.3057,
+      "step": 1554
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6385201066418015e-05,
+      "loss": 1.4583,
+      "step": 1555
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6380032314324928e-05,
+      "loss": 1.5767,
+      "step": 1556
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.637486068604883e-05,
+      "loss": 1.3602,
+      "step": 1557
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6369686183921138e-05,
+      "loss": 1.396,
+      "step": 1558
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6364508810274574e-05,
+      "loss": 1.385,
+      "step": 1559
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6359328567443148e-05,
+      "loss": 1.5199,
+      "step": 1560
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6354145457762164e-05,
+      "loss": 1.5075,
+      "step": 1561
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6348959483568226e-05,
+      "loss": 1.4635,
+      "step": 1562
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.634377064719922e-05,
+      "loss": 1.3497,
+      "step": 1563
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6338578950994325e-05,
+      "loss": 1.377,
+      "step": 1564
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6333384397294012e-05,
+      "loss": 1.3468,
+      "step": 1565
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6328186988440035e-05,
+      "loss": 1.3999,
+      "step": 1566
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6322986726775446e-05,
+      "loss": 1.2494,
+      "step": 1567
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6317783614644564e-05,
+      "loss": 1.3397,
+      "step": 1568
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6312577654393017e-05,
+      "loss": 1.45,
+      "step": 1569
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6307368848367697e-05,
+      "loss": 1.4557,
+      "step": 1570
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6302157198916787e-05,
+      "loss": 1.4798,
+      "step": 1571
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.629694270838975e-05,
+      "loss": 1.376,
+      "step": 1572
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6291725379137335e-05,
+      "loss": 1.2762,
+      "step": 1573
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6286505213511562e-05,
+      "loss": 1.5616,
+      "step": 1574
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.628128221386574e-05,
+      "loss": 1.344,
+      "step": 1575
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6276056382554443e-05,
+      "loss": 1.3232,
+      "step": 1576
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6270827721933537e-05,
+      "loss": 1.2393,
+      "step": 1577
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6265596234360147e-05,
+      "loss": 1.443,
+      "step": 1578
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.626036192219269e-05,
+      "loss": 1.649,
+      "step": 1579
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6255124787790838e-05,
+      "loss": 1.4725,
+      "step": 1580
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6249884833515546e-05,
+      "loss": 1.5928,
+      "step": 1581
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.624464206172904e-05,
+      "loss": 1.5142,
+      "step": 1582
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.623939647479481e-05,
+      "loss": 1.5012,
+      "step": 1583
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6234148075077628e-05,
+      "loss": 1.3933,
+      "step": 1584
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.622889686494352e-05,
+      "loss": 1.2363,
+      "step": 1585
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6223642846759782e-05,
+      "loss": 1.4977,
+      "step": 1586
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6218386022894988e-05,
+      "loss": 1.3015,
+      "step": 1587
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6213126395718956e-05,
+      "loss": 1.4114,
+      "step": 1588
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6207863967602788e-05,
+      "loss": 1.4536,
+      "step": 1589
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6202598740918832e-05,
+      "loss": 1.2023,
+      "step": 1590
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6197330718040712e-05,
+      "loss": 1.4572,
+      "step": 1591
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6192059901343305e-05,
+      "loss": 1.4064,
+      "step": 1592
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.618678629320275e-05,
+      "loss": 1.4302,
+      "step": 1593
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.618150989599643e-05,
+      "loss": 1.418,
+      "step": 1594
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6176230712103013e-05,
+      "loss": 1.3095,
+      "step": 1595
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6170948743902403e-05,
+      "loss": 1.315,
+      "step": 1596
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6165663993775766e-05,
+      "loss": 1.2929,
+      "step": 1597
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.616037646410552e-05,
+      "loss": 1.3194,
+      "step": 1598
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.615508615727533e-05,
+      "loss": 1.4594,
+      "step": 1599
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6149793075670136e-05,
+      "loss": 1.3729,
+      "step": 1600
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6144497221676098e-05,
+      "loss": 1.4637,
+      "step": 1601
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.613919859768064e-05,
+      "loss": 1.3131,
+      "step": 1602
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.613389720607245e-05,
+      "loss": 1.6196,
+      "step": 1603
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.612859304924143e-05,
+      "loss": 1.241,
+      "step": 1604
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6123286129578765e-05,
+      "loss": 1.4954,
+      "step": 1605
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6117976449476856e-05,
+      "loss": 1.335,
+      "step": 1606
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6112664011329364e-05,
+      "loss": 1.3101,
+      "step": 1607
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.610734881753119e-05,
+      "loss": 1.4327,
+      "step": 1608
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.610203087047848e-05,
+      "loss": 1.2698,
+      "step": 1609
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6096710172568613e-05,
+      "loss": 1.2436,
+      "step": 1610
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6091386726200216e-05,
+      "loss": 1.472,
+      "step": 1611
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6086060533773152e-05,
+      "loss": 1.5873,
+      "step": 1612
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6080731597688523e-05,
+      "loss": 1.3379,
+      "step": 1613
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6075399920348666e-05,
+      "loss": 1.4557,
+      "step": 1614
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6070065504157152e-05,
+      "loss": 1.3537,
+      "step": 1615
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.60647283515188e-05,
+      "loss": 1.2663,
+      "step": 1616
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.605938846483964e-05,
+      "loss": 1.554,
+      "step": 1617
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6054045846526956e-05,
+      "loss": 1.4093,
+      "step": 1618
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6048700498989248e-05,
+      "loss": 1.4403,
+      "step": 1619
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.604335242463626e-05,
+      "loss": 1.2583,
+      "step": 1620
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.603800162587895e-05,
+      "loss": 1.2933,
+      "step": 1621
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.603264810512952e-05,
+      "loss": 1.1738,
+      "step": 1622
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.602729186480138e-05,
+      "loss": 1.524,
+      "step": 1623
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6021932907309188e-05,
+      "loss": 1.3831,
+      "step": 1624
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6016571235068813e-05,
+      "loss": 1.2214,
+      "step": 1625
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6011206850497346e-05,
+      "loss": 1.3866,
+      "step": 1626
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6005839756013113e-05,
+      "loss": 1.5282,
+      "step": 1627
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.600046995403565e-05,
+      "loss": 1.1915,
+      "step": 1628
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.599509744698572e-05,
+      "loss": 1.3527,
+      "step": 1629
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5989722237285307e-05,
+      "loss": 1.5028,
+      "step": 1630
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.59843443273576e-05,
+      "loss": 1.2722,
+      "step": 1631
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5978963719627025e-05,
+      "loss": 1.3367,
+      "step": 1632
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5973580416519207e-05,
+      "loss": 1.5668,
+      "step": 1633
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5968194420460996e-05,
+      "loss": 1.3497,
+      "step": 1634
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5962805733880455e-05,
+      "loss": 1.4342,
+      "step": 1635
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5957414359206858e-05,
+      "loss": 1.1661,
+      "step": 1636
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.595202029887069e-05,
+      "loss": 1.3772,
+      "step": 1637
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.594662355530365e-05,
+      "loss": 1.3809,
+      "step": 1638
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5941224130938637e-05,
+      "loss": 1.4543,
+      "step": 1639
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5935822028209776e-05,
+      "loss": 1.4533,
+      "step": 1640
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5930417249552385e-05,
+      "loss": 1.5114,
+      "step": 1641
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5925009797402995e-05,
+      "loss": 1.45,
+      "step": 1642
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5919599674199337e-05,
+      "loss": 1.3144,
+      "step": 1643
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.591418688238035e-05,
+      "loss": 1.0736,
+      "step": 1644
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5908771424386173e-05,
+      "loss": 1.3008,
+      "step": 1645
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.590335330265816e-05,
+      "loss": 1.3119,
+      "step": 1646
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5897932519638836e-05,
+      "loss": 1.2272,
+      "step": 1647
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5892509077771962e-05,
+      "loss": 1.2834,
+      "step": 1648
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5887082979502473e-05,
+      "loss": 1.3102,
+      "step": 1649
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.588165422727651e-05,
+      "loss": 1.139,
+      "step": 1650
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5876222823541406e-05,
+      "loss": 1.0236,
+      "step": 1651
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.58707887707457e-05,
+      "loss": 1.2255,
+      "step": 1652
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.586535207133911e-05,
+      "loss": 1.2757,
+      "step": 1653
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.585991272777256e-05,
+      "loss": 1.1225,
+      "step": 1654
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5854470742498165e-05,
+      "loss": 1.0651,
+      "step": 1655
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5849026117969217e-05,
+      "loss": 1.2055,
+      "step": 1656
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.584357885664021e-05,
+      "loss": 1.0497,
+      "step": 1657
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.583812896096683e-05,
+      "loss": 1.2941,
+      "step": 1658
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5832676433405945e-05,
+      "loss": 1.0475,
+      "step": 1659
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.58272212764156e-05,
+      "loss": 1.2273,
+      "step": 1660
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5821763492455046e-05,
+      "loss": 1.0508,
+      "step": 1661
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.58163030839847e-05,
+      "loss": 1.1571,
+      "step": 1662
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5810840053466174e-05,
+      "loss": 1.1559,
+      "step": 1663
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.580537440336225e-05,
+      "loss": 1.2403,
+      "step": 1664
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5799906136136908e-05,
+      "loss": 1.3124,
+      "step": 1665
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5794435254255294e-05,
+      "loss": 1.0199,
+      "step": 1666
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5788961760183733e-05,
+      "loss": 1.2628,
+      "step": 1667
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5783485656389735e-05,
+      "loss": 1.0911,
+      "step": 1668
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.577800694534198e-05,
+      "loss": 1.0685,
+      "step": 1669
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5772525629510336e-05,
+      "loss": 1.2159,
+      "step": 1670
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5767041711365824e-05,
+      "loss": 1.1649,
+      "step": 1671
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5761555193380656e-05,
+      "loss": 1.0359,
+      "step": 1672
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5756066078028204e-05,
+      "loss": 1.2901,
+      "step": 1673
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5750574367783024e-05,
+      "loss": 1.2533,
+      "step": 1674
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.574508006512083e-05,
+      "loss": 1.1389,
+      "step": 1675
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5739583172518508e-05,
+      "loss": 1.4486,
+      "step": 1676
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.573408369245412e-05,
+      "loss": 1.3568,
+      "step": 1677
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5728581627406877e-05,
+      "loss": 1.2713,
+      "step": 1678
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5723076979857173e-05,
+      "loss": 1.1906,
+      "step": 1679
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5717569752286554e-05,
+      "loss": 1.0598,
+      "step": 1680
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5712059947177737e-05,
+      "loss": 1.2987,
+      "step": 1681
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5706547567014597e-05,
+      "loss": 1.3235,
+      "step": 1682
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.570103261428217e-05,
+      "loss": 1.4125,
+      "step": 1683
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5695515091466653e-05,
+      "loss": 1.152,
+      "step": 1684
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.56899950010554e-05,
+      "loss": 0.9682,
+      "step": 1685
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5684472345536925e-05,
+      "loss": 1.0534,
+      "step": 1686
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5678947127400896e-05,
+      "loss": 1.1776,
+      "step": 1687
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5673419349138138e-05,
+      "loss": 1.2346,
+      "step": 1688
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.566788901324063e-05,
+      "loss": 1.166,
+      "step": 1689
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.56623561222015e-05,
+      "loss": 1.0464,
+      "step": 1690
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5656820678515035e-05,
+      "loss": 1.1854,
+      "step": 1691
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5651282684676665e-05,
+      "loss": 0.9499,
+      "step": 1692
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5645742143182974e-05,
+      "loss": 1.1744,
+      "step": 1693
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.56401990565317e-05,
+      "loss": 1.2945,
+      "step": 1694
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5634653427221717e-05,
+      "loss": 0.9753,
+      "step": 1695
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.562910525775305e-05,
+      "loss": 1.3807,
+      "step": 1696
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5623554550626874e-05,
+      "loss": 1.3427,
+      "step": 1697
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.56180013083455e-05,
+      "loss": 1.2128,
+      "step": 1698
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5612445533412384e-05,
+      "loss": 1.2347,
+      "step": 1699
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5606887228332136e-05,
+      "loss": 1.415,
+      "step": 1700
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5601326395610486e-05,
+      "loss": 1.1894,
+      "step": 1701
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.559576303775431e-05,
+      "loss": 1.1368,
+      "step": 1702
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.559019715727164e-05,
+      "loss": 1.0627,
+      "step": 1703
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.558462875667162e-05,
+      "loss": 1.2814,
+      "step": 1704
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5579057838464545e-05,
+      "loss": 1.1955,
+      "step": 1705
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5573484405161844e-05,
+      "loss": 1.0746,
+      "step": 1706
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5567908459276073e-05,
+      "loss": 1.2386,
+      "step": 1707
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5562330003320925e-05,
+      "loss": 1.0176,
+      "step": 1708
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.555674903981122e-05,
+      "loss": 1.3593,
+      "step": 1709
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.555116557126292e-05,
+      "loss": 1.2904,
+      "step": 1710
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.554557960019311e-05,
+      "loss": 1.2063,
+      "step": 1711
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5539991129119997e-05,
+      "loss": 1.2305,
+      "step": 1712
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.553440016056292e-05,
+      "loss": 1.3646,
+      "step": 1713
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5528806697042355e-05,
+      "loss": 1.1016,
+      "step": 1714
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.552321074107987e-05,
+      "loss": 1.1618,
+      "step": 1715
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.55176122951982e-05,
+      "loss": 1.2075,
+      "step": 1716
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.551201136192117e-05,
+      "loss": 1.102,
+      "step": 1717
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.550640794377374e-05,
+      "loss": 0.9726,
+      "step": 1718
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5500802043281986e-05,
+      "loss": 1.0649,
+      "step": 1719
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.549519366297311e-05,
+      "loss": 1.2003,
+      "step": 1720
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.548958280537542e-05,
+      "loss": 1.2634,
+      "step": 1721
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5483969473018344e-05,
+      "loss": 1.1525,
+      "step": 1722
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.547835366843244e-05,
+      "loss": 1.2327,
+      "step": 1723
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5472735394149364e-05,
+      "loss": 1.1237,
+      "step": 1724
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.546711465270189e-05,
+      "loss": 1.319,
+      "step": 1725
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5461491446623905e-05,
+      "loss": 1.1545,
+      "step": 1726
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.545586577845041e-05,
+      "loss": 0.9929,
+      "step": 1727
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.545023765071751e-05,
+      "loss": 1.3395,
+      "step": 1728
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.544460706596242e-05,
+      "loss": 1.0328,
+      "step": 1729
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.543897402672347e-05,
+      "loss": 1.196,
+      "step": 1730
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.543333853554009e-05,
+      "loss": 1.2567,
+      "step": 1731
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5427700594952813e-05,
+      "loss": 1.0959,
+      "step": 1732
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.542206020750328e-05,
+      "loss": 1.1194,
+      "step": 1733
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5416417375734237e-05,
+      "loss": 1.1876,
+      "step": 1734
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5410772102189522e-05,
+      "loss": 1.1548,
+      "step": 1735
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5405124389414095e-05,
+      "loss": 1.3074,
+      "step": 1736
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5399474239953995e-05,
+      "loss": 1.3138,
+      "step": 1737
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5393821656356363e-05,
+      "loss": 1.017,
+      "step": 1738
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5388166641169443e-05,
+      "loss": 1.2037,
+      "step": 1739
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.538250919694258e-05,
+      "loss": 1.1733,
+      "step": 1740
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5376849326226194e-05,
+      "loss": 1.1183,
+      "step": 1741
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.537118703157182e-05,
+      "loss": 1.1699,
+      "step": 1742
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.536552231553208e-05,
+      "loss": 1.1462,
+      "step": 1743
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5359855180660678e-05,
+      "loss": 1.0886,
+      "step": 1744
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5354185629512418e-05,
+      "loss": 1.1101,
+      "step": 1745
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5348513664643195e-05,
+      "loss": 1.2181,
+      "step": 1746
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.534283928860998e-05,
+      "loss": 1.2041,
+      "step": 1747
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5337162503970853e-05,
+      "loss": 1.2675,
+      "step": 1748
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.533148331328495e-05,
+      "loss": 1.3349,
+      "step": 1749
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.532580171911252e-05,
+      "loss": 1.0624,
+      "step": 1750
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.5320117724014875e-05,
+      "loss": 1.3494,
+      "step": 1751
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.5314431330554423e-05,
+      "loss": 1.2622,
+      "step": 1752
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.5308742541294643e-05,
+      "loss": 1.2007,
+      "step": 1753
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.5303051358800095e-05,
+      "loss": 1.1426,
+      "step": 1754
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.5297357785636434e-05,
+      "loss": 1.1185,
+      "step": 1755
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.5291661824370372e-05,
+      "loss": 1.2871,
+      "step": 1756
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.52859634775697e-05,
+      "loss": 1.3722,
+      "step": 1757
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.5280262747803305e-05,
+      "loss": 1.2462,
+      "step": 1758
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.5274559637641123e-05,
+      "loss": 1.0817,
+      "step": 1759
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.5268854149654177e-05,
+      "loss": 1.3534,
+      "step": 1760
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.526314628641456e-05,
+      "loss": 1.2834,
+      "step": 1761
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.5257436050495431e-05,
+      "loss": 1.2824,
+      "step": 1762
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.5251723444471025e-05,
+      "loss": 1.1297,
+      "step": 1763
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.524600847091664e-05,
+      "loss": 1.0708,
+      "step": 1764
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.5240291132408645e-05,
+      "loss": 1.0759,
+      "step": 1765
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.5234571431524474e-05,
+      "loss": 1.2308,
+      "step": 1766
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.5228849370842625e-05,
+      "loss": 1.1799,
+      "step": 1767
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.5223124952942663e-05,
+      "loss": 1.2048,
+      "step": 1768
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.521739818040521e-05,
+      "loss": 1.0422,
+      "step": 1769
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.5211669055811955e-05,
+      "loss": 1.2704,
+      "step": 1770
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.520593758174565e-05,
+      "loss": 1.3143,
+      "step": 1771
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.5200203760790091e-05,
+      "loss": 1.0029,
+      "step": 1772
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.5194467595530149e-05,
+      "loss": 1.5197,
+      "step": 1773
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.5188729088551745e-05,
+      "loss": 1.3537,
+      "step": 1774
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.5182988242441857e-05,
+      "loss": 1.086,
+      "step": 1775
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.5177245059788513e-05,
+      "loss": 1.2198,
+      "step": 1776
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.5171499543180797e-05,
+      "loss": 1.0903,
+      "step": 1777
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.5165751695208852e-05,
+      "loss": 1.2782,
+      "step": 1778
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.5160001518463863e-05,
+      "loss": 1.2666,
+      "step": 1779
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.5154249015538063e-05,
+      "loss": 1.311,
+      "step": 1780
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.5148494189024743e-05,
+      "loss": 1.2808,
+      "step": 1781
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.514273704151824e-05,
+      "loss": 1.231,
+      "step": 1782
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.5136977575613924e-05,
+      "loss": 1.251,
+      "step": 1783
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.513121579390823e-05,
+      "loss": 1.1312,
+      "step": 1784
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.5125451698998626e-05,
+      "loss": 1.3436,
+      "step": 1785
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.5119685293483622e-05,
+      "loss": 1.1263,
+      "step": 1786
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.5113916579962773e-05,
+      "loss": 1.0087,
+      "step": 1787
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.5108145561036669e-05,
+      "loss": 1.4066,
+      "step": 1788
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.510237223930695e-05,
+      "loss": 1.1023,
+      "step": 1789
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.5096596617376285e-05,
+      "loss": 1.0587,
+      "step": 1790
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.5090818697848384e-05,
+      "loss": 1.2144,
+      "step": 1791
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.5085038483327987e-05,
+      "loss": 1.0801,
+      "step": 1792
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.5079255976420877e-05,
+      "loss": 1.208,
+      "step": 1793
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.5073471179733869e-05,
+      "loss": 1.1569,
+      "step": 1794
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.5067684095874803e-05,
+      "loss": 1.1094,
+      "step": 1795
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.5061894727452555e-05,
+      "loss": 1.0205,
+      "step": 1796
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.5056103077077033e-05,
+      "loss": 1.4093,
+      "step": 1797
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.5050309147359174e-05,
+      "loss": 1.2054,
+      "step": 1798
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.5044512940910934e-05,
+      "loss": 1.2571,
+      "step": 1799
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.5038714460345302e-05,
+      "loss": 1.243,
+      "step": 1800
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.50329137082763e-05,
+      "loss": 1.1659,
+      "step": 1801
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.5027110687318954e-05,
+      "loss": 1.3058,
+      "step": 1802
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.5021305400089333e-05,
+      "loss": 1.2172,
+      "step": 1803
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.5015497849204512e-05,
+      "loss": 1.2713,
+      "step": 1804
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.5009688037282604e-05,
+      "loss": 1.2823,
+      "step": 1805
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.5003875966942718e-05,
+      "loss": 1.0653,
+      "step": 1806
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4998061640805007e-05,
+      "loss": 1.1762,
+      "step": 1807
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4992245061490618e-05,
+      "loss": 1.3018,
+      "step": 1808
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4986426231621724e-05,
+      "loss": 1.1146,
+      "step": 1809
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4980605153821521e-05,
+      "loss": 1.0583,
+      "step": 1810
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4974781830714203e-05,
+      "loss": 1.2242,
+      "step": 1811
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4968956264924984e-05,
+      "loss": 1.2077,
+      "step": 1812
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4963128459080088e-05,
+      "loss": 1.0819,
+      "step": 1813
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4957298415806749e-05,
+      "loss": 1.3546,
+      "step": 1814
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4951466137733206e-05,
+      "loss": 1.2273,
+      "step": 1815
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4945631627488715e-05,
+      "loss": 1.2737,
+      "step": 1816
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.493979488770353e-05,
+      "loss": 1.4581,
+      "step": 1817
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4933955921008907e-05,
+      "loss": 1.4003,
+      "step": 1818
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.492811473003712e-05,
+      "loss": 1.1177,
+      "step": 1819
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4922271317421431e-05,
+      "loss": 0.9393,
+      "step": 1820
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4916425685796113e-05,
+      "loss": 1.0412,
+      "step": 1821
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4910577837796429e-05,
+      "loss": 0.9391,
+      "step": 1822
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4904727776058656e-05,
+      "loss": 1.2543,
+      "step": 1823
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4898875503220059e-05,
+      "loss": 1.3617,
+      "step": 1824
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4893021021918897e-05,
+      "loss": 1.3303,
+      "step": 1825
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4887164334794433e-05,
+      "loss": 1.0862,
+      "step": 1826
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.488130544448692e-05,
+      "loss": 1.1649,
+      "step": 1827
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4875444353637608e-05,
+      "loss": 1.0991,
+      "step": 1828
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.486958106488873e-05,
+      "loss": 1.2602,
+      "step": 1829
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4863715580883522e-05,
+      "loss": 1.0196,
+      "step": 1830
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4857847904266198e-05,
+      "loss": 1.054,
+      "step": 1831
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.485197803768197e-05,
+      "loss": 0.9955,
+      "step": 1832
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.484610598377703e-05,
+      "loss": 1.0817,
+      "step": 1833
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4840231745198564e-05,
+      "loss": 1.282,
+      "step": 1834
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.483435532459473e-05,
+      "loss": 1.3281,
+      "step": 1835
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4828476724614684e-05,
+      "loss": 1.1753,
+      "step": 1836
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.482259594790856e-05,
+      "loss": 1.0709,
+      "step": 1837
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4816712997127464e-05,
+      "loss": 1.2132,
+      "step": 1838
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4810827874923495e-05,
+      "loss": 1.0938,
+      "step": 1839
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4804940583949727e-05,
+      "loss": 1.2847,
+      "step": 1840
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4799051126860207e-05,
+      "loss": 1.1984,
+      "step": 1841
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.479315950630996e-05,
+      "loss": 1.0925,
+      "step": 1842
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.478726572495499e-05,
+      "loss": 1.2929,
+      "step": 1843
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4781369785452277e-05,
+      "loss": 1.1371,
+      "step": 1844
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4775471690459768e-05,
+      "loss": 1.0288,
+      "step": 1845
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4769571442636381e-05,
+      "loss": 1.2399,
+      "step": 1846
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4763669044642007e-05,
+      "loss": 1.1037,
+      "step": 1847
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.475776449913751e-05,
+      "loss": 1.099,
+      "step": 1848
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4751857808784718e-05,
+      "loss": 1.1975,
+      "step": 1849
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4745948976246424e-05,
+      "loss": 1.2215,
+      "step": 1850
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4740038004186393e-05,
+      "loss": 1.0093,
+      "step": 1851
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4734124895269348e-05,
+      "loss": 1.378,
+      "step": 1852
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4728209652160983e-05,
+      "loss": 1.0601,
+      "step": 1853
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.472229227752794e-05,
+      "loss": 1.0787,
+      "step": 1854
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4716372774037842e-05,
+      "loss": 1.1764,
+      "step": 1855
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4710451144359254e-05,
+      "loss": 1.2725,
+      "step": 1856
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4704527391161709e-05,
+      "loss": 1.2457,
+      "step": 1857
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4698601517115696e-05,
+      "loss": 1.1423,
+      "step": 1858
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4692673524892655e-05,
+      "loss": 0.9695,
+      "step": 1859
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4686743417164987e-05,
+      "loss": 1.1832,
+      "step": 1860
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4680811196606046e-05,
+      "loss": 1.2145,
+      "step": 1861
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4674876865890135e-05,
+      "loss": 1.2635,
+      "step": 1862
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4668940427692511e-05,
+      "loss": 1.2779,
+      "step": 1863
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4663001884689379e-05,
+      "loss": 1.2144,
+      "step": 1864
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4657061239557898e-05,
+      "loss": 1.2195,
+      "step": 1865
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4651118494976161e-05,
+      "loss": 1.2884,
+      "step": 1866
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.464517365362323e-05,
+      "loss": 1.3029,
+      "step": 1867
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4639226718179092e-05,
+      "loss": 1.138,
+      "step": 1868
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.463327769132469e-05,
+      "loss": 1.0732,
+      "step": 1869
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.46273265757419e-05,
+      "loss": 1.151,
+      "step": 1870
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4621373374113548e-05,
+      "loss": 0.9851,
+      "step": 1871
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4615418089123399e-05,
+      "loss": 1.3317,
+      "step": 1872
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.460946072345615e-05,
+      "loss": 1.2028,
+      "step": 1873
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.460350127979745e-05,
+      "loss": 1.3085,
+      "step": 1874
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4597539760833867e-05,
+      "loss": 1.1485,
+      "step": 1875
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4591576169252926e-05,
+      "loss": 1.2948,
+      "step": 1876
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.458561050774306e-05,
+      "loss": 1.2383,
+      "step": 1877
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4579642778993659e-05,
+      "loss": 1.2241,
+      "step": 1878
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4573672985695035e-05,
+      "loss": 1.0633,
+      "step": 1879
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4567701130538425e-05,
+      "loss": 1.2214,
+      "step": 1880
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.456172721621601e-05,
+      "loss": 1.4536,
+      "step": 1881
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4555751245420884e-05,
+      "loss": 1.2873,
+      "step": 1882
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4549773220847079e-05,
+      "loss": 1.3289,
+      "step": 1883
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4543793145189549e-05,
+      "loss": 1.3319,
+      "step": 1884
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4537811021144168e-05,
+      "loss": 1.1753,
+      "step": 1885
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.453182685140774e-05,
+      "loss": 1.1917,
+      "step": 1886
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4525840638677994e-05,
+      "loss": 1.0893,
+      "step": 1887
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4519852385653572e-05,
+      "loss": 1.0934,
+      "step": 1888
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4513862095034037e-05,
+      "loss": 1.0533,
+      "step": 1889
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4507869769519869e-05,
+      "loss": 1.0928,
+      "step": 1890
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.450187541181248e-05,
+      "loss": 1.2456,
+      "step": 1891
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4495879024614178e-05,
+      "loss": 1.2253,
+      "step": 1892
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4489880610628198e-05,
+      "loss": 1.407,
+      "step": 1893
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4483880172558684e-05,
+      "loss": 1.4026,
+      "step": 1894
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4477877713110697e-05,
+      "loss": 1.0005,
+      "step": 1895
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4471873234990206e-05,
+      "loss": 1.3185,
+      "step": 1896
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4465866740904086e-05,
+      "loss": 1.2993,
+      "step": 1897
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4459858233560132e-05,
+      "loss": 1.0783,
+      "step": 1898
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4453847715667032e-05,
+      "loss": 1.1248,
+      "step": 1899
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4447835189934394e-05,
+      "loss": 1.4129,
+      "step": 1900
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4441820659072725e-05,
+      "loss": 1.031,
+      "step": 1901
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4435804125793431e-05,
+      "loss": 1.0454,
+      "step": 1902
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4429785592808831e-05,
+      "loss": 1.1745,
+      "step": 1903
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4423765062832142e-05,
+      "loss": 1.1005,
+      "step": 1904
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4417742538577476e-05,
+      "loss": 1.1742,
+      "step": 1905
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4411718022759847e-05,
+      "loss": 1.0946,
+      "step": 1906
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4405691518095173e-05,
+      "loss": 1.2304,
+      "step": 1907
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4399663027300257e-05,
+      "loss": 1.3028,
+      "step": 1908
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4393632553092807e-05,
+      "loss": 1.2162,
+      "step": 1909
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.438760009819142e-05,
+      "loss": 1.2239,
+      "step": 1910
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4381565665315588e-05,
+      "loss": 1.2678,
+      "step": 1911
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4375529257185694e-05,
+      "loss": 1.1802,
+      "step": 1912
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4369490876523015e-05,
+      "loss": 1.3378,
+      "step": 1913
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.4363450526049709e-05,
+      "loss": 1.1258,
+      "step": 1914
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.435740820848883e-05,
+      "loss": 1.0813,
+      "step": 1915
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.435136392656432e-05,
+      "loss": 1.1506,
+      "step": 1916
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.434531768300099e-05,
+      "loss": 1.0578,
+      "step": 1917
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.4339269480524561e-05,
+      "loss": 1.163,
+      "step": 1918
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.4333219321861616e-05,
+      "loss": 1.2235,
+      "step": 1919
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.4327167209739629e-05,
+      "loss": 1.0708,
+      "step": 1920
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.4321113146886953e-05,
+      "loss": 1.0935,
+      "step": 1921
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.4315057136032824e-05,
+      "loss": 1.0403,
+      "step": 1922
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.430899917990735e-05,
+      "loss": 1.2254,
+      "step": 1923
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.4302939281241522e-05,
+      "loss": 1.293,
+      "step": 1924
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.4296877442767194e-05,
+      "loss": 1.3429,
+      "step": 1925
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.4290813667217117e-05,
+      "loss": 1.2502,
+      "step": 1926
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.4284747957324893e-05,
+      "loss": 1.0905,
+      "step": 1927
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.427868031582501e-05,
+      "loss": 1.2244,
+      "step": 1928
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.4272610745452818e-05,
+      "loss": 1.2131,
+      "step": 1929
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.4266539248944547e-05,
+      "loss": 1.3635,
+      "step": 1930
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.4260465829037283e-05,
+      "loss": 1.1078,
+      "step": 1931
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.4254390488468987e-05,
+      "loss": 1.228,
+      "step": 1932
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.4248313229978484e-05,
+      "loss": 1.2348,
+      "step": 1933
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.4242234056305463e-05,
+      "loss": 0.9427,
+      "step": 1934
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.423615297019048e-05,
+      "loss": 1.1989,
+      "step": 1935
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.4230069974374945e-05,
+      "loss": 1.1877,
+      "step": 1936
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.4223985071601139e-05,
+      "loss": 1.3003,
+      "step": 1937
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.4217898264612191e-05,
+      "loss": 1.3912,
+      "step": 1938
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.4211809556152104e-05,
+      "loss": 1.1203,
+      "step": 1939
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.4205718948965723e-05,
+      "loss": 1.355,
+      "step": 1940
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.4199626445798753e-05,
+      "loss": 1.2411,
+      "step": 1941
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.4193532049397762e-05,
+      "loss": 1.3325,
+      "step": 1942
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.4187435762510167e-05,
+      "loss": 1.211,
+      "step": 1943
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.4181337587884227e-05,
+      "loss": 1.2129,
+      "step": 1944
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.4175237528269066e-05,
+      "loss": 1.1728,
+      "step": 1945
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.4169135586414654e-05,
+      "loss": 1.2395,
+      "step": 1946
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.4163031765071806e-05,
+      "loss": 1.3407,
+      "step": 1947
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.4156926066992184e-05,
+      "loss": 1.2577,
+      "step": 1948
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.4150818494928303e-05,
+      "loss": 1.332,
+      "step": 1949
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.4144709051633516e-05,
+      "loss": 0.9956,
+      "step": 1950
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.4138597739862021e-05,
+      "loss": 1.2109,
+      "step": 1951
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.413248456236886e-05,
+      "loss": 1.035,
+      "step": 1952
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.4126369521909916e-05,
+      "loss": 1.2527,
+      "step": 1953
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.4120252621241905e-05,
+      "loss": 1.2027,
+      "step": 1954
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.4114133863122399e-05,
+      "loss": 1.2015,
+      "step": 1955
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.4108013250309784e-05,
+      "loss": 1.062,
+      "step": 1956
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.4101890785563304e-05,
+      "loss": 1.4457,
+      "step": 1957
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.4095766471643022e-05,
+      "loss": 1.3294,
+      "step": 1958
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.4089640311309842e-05,
+      "loss": 1.2185,
+      "step": 1959
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.40835123073255e-05,
+      "loss": 0.9887,
+      "step": 1960
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.407738246245256e-05,
+      "loss": 1.1853,
+      "step": 1961
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.407125077945442e-05,
+      "loss": 1.4654,
+      "step": 1962
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.4065117261095304e-05,
+      "loss": 1.1319,
+      "step": 1963
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.4058981910140263e-05,
+      "loss": 1.2954,
+      "step": 1964
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.4052844729355176e-05,
+      "loss": 1.1022,
+      "step": 1965
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.4046705721506748e-05,
+      "loss": 1.4003,
+      "step": 1966
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.40405648893625e-05,
+      "loss": 1.1,
+      "step": 1967
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.4034422235690786e-05,
+      "loss": 1.0683,
+      "step": 1968
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.4028277763260778e-05,
+      "loss": 1.3299,
+      "step": 1969
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.4022131474842463e-05,
+      "loss": 1.2877,
+      "step": 1970
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.401598337320665e-05,
+      "loss": 1.0321,
+      "step": 1971
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.4009833461124965e-05,
+      "loss": 1.4612,
+      "step": 1972
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.4003681741369852e-05,
+      "loss": 1.184,
+      "step": 1973
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3997528216714566e-05,
+      "loss": 1.2734,
+      "step": 1974
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3991372889933184e-05,
+      "loss": 1.2992,
+      "step": 1975
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3985215763800582e-05,
+      "loss": 1.1796,
+      "step": 1976
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3979056841092457e-05,
+      "loss": 1.1665,
+      "step": 1977
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3972896124585318e-05,
+      "loss": 1.091,
+      "step": 1978
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3966733617056477e-05,
+      "loss": 1.0773,
+      "step": 1979
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3960569321284044e-05,
+      "loss": 1.2336,
+      "step": 1980
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3954403240046958e-05,
+      "loss": 1.1859,
+      "step": 1981
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.394823537612495e-05,
+      "loss": 1.297,
+      "step": 1982
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3942065732298549e-05,
+      "loss": 1.3955,
+      "step": 1983
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3935894311349096e-05,
+      "loss": 1.1309,
+      "step": 1984
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.392972111605873e-05,
+      "loss": 1.1994,
+      "step": 1985
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3923546149210389e-05,
+      "loss": 1.0352,
+      "step": 1986
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3917369413587816e-05,
+      "loss": 1.2486,
+      "step": 1987
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3911190911975533e-05,
+      "loss": 1.4564,
+      "step": 1988
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3905010647158888e-05,
+      "loss": 1.0112,
+      "step": 1989
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.389882862192399e-05,
+      "loss": 1.2672,
+      "step": 1990
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3892644839057771e-05,
+      "loss": 1.3008,
+      "step": 1991
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3886459301347934e-05,
+      "loss": 0.9989,
+      "step": 1992
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3880272011582987e-05,
+      "loss": 1.4126,
+      "step": 1993
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3874082972552226e-05,
+      "loss": 1.4946,
+      "step": 1994
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3867892187045722e-05,
+      "loss": 1.2388,
+      "step": 1995
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3861699657854351e-05,
+      "loss": 1.1808,
+      "step": 1996
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3855505387769767e-05,
+      "loss": 1.1693,
+      "step": 1997
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.384930937958441e-05,
+      "loss": 1.1548,
+      "step": 1998
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3843111636091499e-05,
+      "loss": 1.204,
+      "step": 1999
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3836912160085043e-05,
+      "loss": 1.1955,
+      "step": 2000
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.383071095435983e-05,
+      "loss": 1.3003,
+      "step": 2001
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3824508021711421e-05,
+      "loss": 1.2171,
+      "step": 2002
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3818303364936164e-05,
+      "loss": 1.059,
+      "step": 2003
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3812096986831178e-05,
+      "loss": 1.2108,
+      "step": 2004
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3805888890194368e-05,
+      "loss": 1.2182,
+      "step": 2005
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.37996790778244e-05,
+      "loss": 1.1763,
+      "step": 2006
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.379346755252072e-05,
+      "loss": 0.9098,
+      "step": 2007
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3787254317083552e-05,
+      "loss": 1.1505,
+      "step": 2008
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3781039374313879e-05,
+      "loss": 1.2112,
+      "step": 2009
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3774822727013463e-05,
+      "loss": 1.1831,
+      "step": 2010
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.376860437798483e-05,
+      "loss": 1.1357,
+      "step": 2011
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.376238433003128e-05,
+      "loss": 1.1943,
+      "step": 2012
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3756162585956868e-05,
+      "loss": 1.2423,
+      "step": 2013
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.374993914856642e-05,
+      "loss": 1.1959,
+      "step": 2014
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3743714020665522e-05,
+      "loss": 1.2757,
+      "step": 2015
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3737487205060528e-05,
+      "loss": 1.3046,
+      "step": 2016
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3731258704558552e-05,
+      "loss": 1.1561,
+      "step": 2017
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.372502852196746e-05,
+      "loss": 1.1862,
+      "step": 2018
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3718796660095878e-05,
+      "loss": 1.3501,
+      "step": 2019
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3712563121753202e-05,
+      "loss": 1.3771,
+      "step": 2020
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.370632790974957e-05,
+      "loss": 1.0229,
+      "step": 2021
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3700091026895876e-05,
+      "loss": 1.0669,
+      "step": 2022
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3693852476003771e-05,
+      "loss": 1.3499,
+      "step": 2023
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3687612259885657e-05,
+      "loss": 1.4938,
+      "step": 2024
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3681370381354686e-05,
+      "loss": 1.1683,
+      "step": 2025
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3675126843224763e-05,
+      "loss": 1.1315,
+      "step": 2026
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3668881648310532e-05,
+      "loss": 1.3982,
+      "step": 2027
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3662634799427397e-05,
+      "loss": 1.2808,
+      "step": 2028
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3656386299391497e-05,
+      "loss": 1.1882,
+      "step": 2029
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.365013615101972e-05,
+      "loss": 1.0497,
+      "step": 2030
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3643884357129693e-05,
+      "loss": 1.1544,
+      "step": 2031
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3637630920539788e-05,
+      "loss": 1.3117,
+      "step": 2032
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3631375844069124e-05,
+      "loss": 1.1025,
+      "step": 2033
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3625119130537545e-05,
+      "loss": 1.1546,
+      "step": 2034
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3618860782765642e-05,
+      "loss": 1.069,
+      "step": 2035
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3612600803574747e-05,
+      "loss": 1.2453,
+      "step": 2036
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3606339195786914e-05,
+      "loss": 1.0729,
+      "step": 2037
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3600075962224946e-05,
+      "loss": 1.1801,
+      "step": 2038
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3593811105712367e-05,
+      "loss": 1.2086,
+      "step": 2039
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3587544629073442e-05,
+      "loss": 1.2389,
+      "step": 2040
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3581276535133157e-05,
+      "loss": 1.1315,
+      "step": 2041
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3575006826717236e-05,
+      "loss": 1.2047,
+      "step": 2042
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3568735506652124e-05,
+      "loss": 1.4608,
+      "step": 2043
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3562462577764999e-05,
+      "loss": 1.3353,
+      "step": 2044
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3556188042883758e-05,
+      "loss": 1.3626,
+      "step": 2045
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3549911904837022e-05,
+      "loss": 1.0906,
+      "step": 2046
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3543634166454144e-05,
+      "loss": 1.3742,
+      "step": 2047
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3537354830565188e-05,
+      "loss": 1.1893,
+      "step": 2048
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3531073900000942e-05,
+      "loss": 1.2246,
+      "step": 2049
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3524791377592914e-05,
+      "loss": 1.3342,
+      "step": 2050
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3518507266173326e-05,
+      "loss": 1.1955,
+      "step": 2051
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3512221568575126e-05,
+      "loss": 1.2288,
+      "step": 2052
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.350593428763196e-05,
+      "loss": 1.1837,
+      "step": 2053
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3499645426178205e-05,
+      "loss": 1.2081,
+      "step": 2054
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3493354987048941e-05,
+      "loss": 0.9495,
+      "step": 2055
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3487062973079964e-05,
+      "loss": 1.0695,
+      "step": 2056
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.348076938710777e-05,
+      "loss": 1.2317,
+      "step": 2057
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3474474231969578e-05,
+      "loss": 1.2005,
+      "step": 2058
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3468177510503306e-05,
+      "loss": 1.2447,
+      "step": 2059
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.346187922554758e-05,
+      "loss": 1.347,
+      "step": 2060
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.345557937994173e-05,
+      "loss": 1.2506,
+      "step": 2061
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3449277976525785e-05,
+      "loss": 1.2922,
+      "step": 2062
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3442975018140491e-05,
+      "loss": 0.9299,
+      "step": 2063
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3436670507627282e-05,
+      "loss": 1.0966,
+      "step": 2064
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.343036444782829e-05,
+      "loss": 1.1562,
+      "step": 2065
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3424056841586355e-05,
+      "loss": 1.3465,
+      "step": 2066
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3417747691745007e-05,
+      "loss": 1.2112,
+      "step": 2067
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3411437001148478e-05,
+      "loss": 1.1478,
+      "step": 2068
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.340512477264169e-05,
+      "loss": 1.2101,
+      "step": 2069
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3398811009070254e-05,
+      "loss": 1.3102,
+      "step": 2070
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3392495713280484e-05,
+      "loss": 1.1991,
+      "step": 2071
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.338617888811938e-05,
+      "loss": 1.1726,
+      "step": 2072
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.337986053643463e-05,
+      "loss": 1.1646,
+      "step": 2073
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3373540661074605e-05,
+      "loss": 1.2363,
+      "step": 2074
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3367219264888375e-05,
+      "loss": 1.1204,
+      "step": 2075
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3360896350725688e-05,
+      "loss": 1.3136,
+      "step": 2076
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3354571921436973e-05,
+      "loss": 1.0803,
+      "step": 2077
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.3348245979873355e-05,
+      "loss": 1.1654,
+      "step": 2078
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.3341918528886626e-05,
+      "loss": 1.1043,
+      "step": 2079
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.3335589571329266e-05,
+      "loss": 1.1951,
+      "step": 2080
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.3329259110054436e-05,
+      "loss": 1.3578,
+      "step": 2081
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.3322927147915966e-05,
+      "loss": 1.1068,
+      "step": 2082
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.3316593687768373e-05,
+      "loss": 1.0123,
+      "step": 2083
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.3310258732466846e-05,
+      "loss": 1.2206,
+      "step": 2084
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.3303922284867243e-05,
+      "loss": 1.3974,
+      "step": 2085
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.3297584347826098e-05,
+      "loss": 1.1985,
+      "step": 2086
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.3291244924200621e-05,
+      "loss": 0.9673,
+      "step": 2087
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.328490401684869e-05,
+      "loss": 1.2878,
+      "step": 2088
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.3278561628628841e-05,
+      "loss": 1.1615,
+      "step": 2089
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.3272217762400295e-05,
+      "loss": 1.1369,
+      "step": 2090
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.3265872421022927e-05,
+      "loss": 1.191,
+      "step": 2091
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.3259525607357285e-05,
+      "loss": 1.0651,
+      "step": 2092
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.325317732426457e-05,
+      "loss": 1.019,
+      "step": 2093
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.3246827574606655e-05,
+      "loss": 1.1033,
+      "step": 2094
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.3240476361246078e-05,
+      "loss": 1.2996,
+      "step": 2095
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.3234123687046019e-05,
+      "loss": 1.0164,
+      "step": 2096
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.3227769554870337e-05,
+      "loss": 1.2083,
+      "step": 2097
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.3221413967583533e-05,
+      "loss": 1.2938,
+      "step": 2098
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.321505692805077e-05,
+      "loss": 1.0848,
+      "step": 2099
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.3208698439137873e-05,
+      "loss": 1.2013,
+      "step": 2100
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.3202338503711305e-05,
+      "loss": 1.1857,
+      "step": 2101
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.319597712463819e-05,
+      "loss": 1.6231,
+      "step": 2102
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.3189614304786307e-05,
+      "loss": 1.1969,
+      "step": 2103
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.3183250047024077e-05,
+      "loss": 1.0983,
+      "step": 2104
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.3176884354220566e-05,
+      "loss": 1.2159,
+      "step": 2105
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.3170517229245498e-05,
+      "loss": 1.253,
+      "step": 2106
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.3164148674969242e-05,
+      "loss": 1.3674,
+      "step": 2107
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.31577786942628e-05,
+      "loss": 1.0396,
+      "step": 2108
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.3151407289997821e-05,
+      "loss": 1.3316,
+      "step": 2109
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.3145034465046604e-05,
+      "loss": 1.1121,
+      "step": 2110
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.3138660222282085e-05,
+      "loss": 1.188,
+      "step": 2111
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.3132284564577828e-05,
+      "loss": 1.0669,
+      "step": 2112
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.3125907494808052e-05,
+      "loss": 1.0432,
+      "step": 2113
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.3119529015847603e-05,
+      "loss": 1.2992,
+      "step": 2114
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.3113149130571962e-05,
+      "loss": 1.1576,
+      "step": 2115
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.3106767841857245e-05,
+      "loss": 1.3185,
+      "step": 2116
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.3100385152580202e-05,
+      "loss": 0.9731,
+      "step": 2117
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.3094001065618216e-05,
+      "loss": 1.2049,
+      "step": 2118
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.3087615583849293e-05,
+      "loss": 1.3477,
+      "step": 2119
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.308122871015208e-05,
+      "loss": 1.1185,
+      "step": 2120
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.3074840447405839e-05,
+      "loss": 1.2086,
+      "step": 2121
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.3068450798490462e-05,
+      "loss": 1.033,
+      "step": 2122
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.3062059766286473e-05,
+      "loss": 1.4609,
+      "step": 2123
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.305566735367501e-05,
+      "loss": 1.2377,
+      "step": 2124
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.3049273563537837e-05,
+      "loss": 1.2208,
+      "step": 2125
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.304287839875734e-05,
+      "loss": 1.3202,
+      "step": 2126
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.3036481862216525e-05,
+      "loss": 1.4078,
+      "step": 2127
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.3030083956799013e-05,
+      "loss": 1.4104,
+      "step": 2128
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.3023684685389047e-05,
+      "loss": 1.2805,
+      "step": 2129
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.3017284050871482e-05,
+      "loss": 1.2929,
+      "step": 2130
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.3010882056131786e-05,
+      "loss": 1.1087,
+      "step": 2131
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.3004478704056048e-05,
+      "loss": 1.0888,
+      "step": 2132
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2998073997530957e-05,
+      "loss": 1.181,
+      "step": 2133
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2991667939443828e-05,
+      "loss": 1.2782,
+      "step": 2134
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.298526053268257e-05,
+      "loss": 1.2598,
+      "step": 2135
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2978851780135711e-05,
+      "loss": 1.2695,
+      "step": 2136
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2972441684692376e-05,
+      "loss": 1.2085,
+      "step": 2137
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2966030249242306e-05,
+      "loss": 1.3088,
+      "step": 2138
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2959617476675842e-05,
+      "loss": 1.2143,
+      "step": 2139
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2953203369883918e-05,
+      "loss": 1.2291,
+      "step": 2140
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2946787931758084e-05,
+      "loss": 1.1215,
+      "step": 2141
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2940371165190488e-05,
+      "loss": 1.2876,
+      "step": 2142
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2933953073073871e-05,
+      "loss": 1.1846,
+      "step": 2143
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2927533658301567e-05,
+      "loss": 1.0214,
+      "step": 2144
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2921112923767519e-05,
+      "loss": 1.3311,
+      "step": 2145
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2914690872366257e-05,
+      "loss": 1.1964,
+      "step": 2146
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2908267506992913e-05,
+      "loss": 1.2744,
+      "step": 2147
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2901842830543193e-05,
+      "loss": 1.183,
+      "step": 2148
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2895416845913418e-05,
+      "loss": 1.0477,
+      "step": 2149
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.288898955600048e-05,
+      "loss": 1.2482,
+      "step": 2150
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2882560963701864e-05,
+      "loss": 1.2886,
+      "step": 2151
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2876131071915652e-05,
+      "loss": 1.1797,
+      "step": 2152
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2869699883540498e-05,
+      "loss": 1.5842,
+      "step": 2153
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.286326740147565e-05,
+      "loss": 1.2937,
+      "step": 2154
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2856833628620937e-05,
+      "loss": 1.1786,
+      "step": 2155
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2850398567876762e-05,
+      "loss": 1.0426,
+      "step": 2156
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2843962222144123e-05,
+      "loss": 1.0637,
+      "step": 2157
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2837524594324587e-05,
+      "loss": 1.083,
+      "step": 2158
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2831085687320301e-05,
+      "loss": 1.1393,
+      "step": 2159
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2824645504033988e-05,
+      "loss": 0.9945,
+      "step": 2160
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.281820404736895e-05,
+      "loss": 1.2315,
+      "step": 2161
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2811761320229059e-05,
+      "loss": 1.2882,
+      "step": 2162
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2805317325518767e-05,
+      "loss": 1.3085,
+      "step": 2163
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.279887206614308e-05,
+      "loss": 0.976,
+      "step": 2164
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2792425545007598e-05,
+      "loss": 1.1241,
+      "step": 2165
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2785977765018474e-05,
+      "loss": 0.9634,
+      "step": 2166
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.277952872908243e-05,
+      "loss": 1.1645,
+      "step": 2167
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2773078440106757e-05,
+      "loss": 1.2553,
+      "step": 2168
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2766626900999316e-05,
+      "loss": 1.3783,
+      "step": 2169
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2760174114668518e-05,
+      "loss": 1.2872,
+      "step": 2170
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2753720084023348e-05,
+      "loss": 1.4713,
+      "step": 2171
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2747264811973348e-05,
+      "loss": 1.3472,
+      "step": 2172
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2740808301428623e-05,
+      "loss": 1.3515,
+      "step": 2173
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2734350555299826e-05,
+      "loss": 1.1213,
+      "step": 2174
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2727891576498182e-05,
+      "loss": 1.0332,
+      "step": 2175
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2721431367935455e-05,
+      "loss": 1.0153,
+      "step": 2176
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2714969932523979e-05,
+      "loss": 1.0336,
+      "step": 2177
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2708507273176636e-05,
+      "loss": 1.3256,
+      "step": 2178
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2702043392806856e-05,
+      "loss": 1.3247,
+      "step": 2179
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2695578294328612e-05,
+      "loss": 1.1693,
+      "step": 2180
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.268911198065645e-05,
+      "loss": 1.1771,
+      "step": 2181
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2682644454705446e-05,
+      "loss": 1.1269,
+      "step": 2182
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2676175719391219e-05,
+      "loss": 1.1953,
+      "step": 2183
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2669705777629946e-05,
+      "loss": 1.2923,
+      "step": 2184
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2663234632338344e-05,
+      "loss": 1.128,
+      "step": 2185
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2656762286433666e-05,
+      "loss": 1.3354,
+      "step": 2186
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2650288742833712e-05,
+      "loss": 0.9728,
+      "step": 2187
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.264381400445682e-05,
+      "loss": 1.3681,
+      "step": 2188
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2637338074221873e-05,
+      "loss": 1.2773,
+      "step": 2189
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2630860955048278e-05,
+      "loss": 1.1472,
+      "step": 2190
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2624382649855993e-05,
+      "loss": 1.1748,
+      "step": 2191
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2617903161565495e-05,
+      "loss": 1.271,
+      "step": 2192
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2611422493097809e-05,
+      "loss": 1.128,
+      "step": 2193
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2604940647374488e-05,
+      "loss": 1.2122,
+      "step": 2194
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2598457627317605e-05,
+      "loss": 1.1853,
+      "step": 2195
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2591973435849778e-05,
+      "loss": 1.3241,
+      "step": 2196
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2585488075894146e-05,
+      "loss": 1.19,
+      "step": 2197
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2579001550374373e-05,
+      "loss": 1.1688,
+      "step": 2198
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2572513862214647e-05,
+      "loss": 1.0954,
+      "step": 2199
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2566025014339686e-05,
+      "loss": 1.0729,
+      "step": 2200
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.255953500967473e-05,
+      "loss": 1.146,
+      "step": 2201
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2553043851145539e-05,
+      "loss": 1.3855,
+      "step": 2202
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2546551541678387e-05,
+      "loss": 0.9791,
+      "step": 2203
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2540058084200078e-05,
+      "loss": 1.1465,
+      "step": 2204
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2533563481637926e-05,
+      "loss": 1.1561,
+      "step": 2205
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2527067736919767e-05,
+      "loss": 1.2074,
+      "step": 2206
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2520570852973941e-05,
+      "loss": 1.1417,
+      "step": 2207
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2514072832729318e-05,
+      "loss": 1.259,
+      "step": 2208
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2507573679115264e-05,
+      "loss": 1.363,
+      "step": 2209
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.250107339506167e-05,
+      "loss": 1.2476,
+      "step": 2210
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.249457198349892e-05,
+      "loss": 1.1475,
+      "step": 2211
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2488069447357922e-05,
+      "loss": 1.4339,
+      "step": 2212
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.248156578957009e-05,
+      "loss": 1.1,
+      "step": 2213
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2475061013067328e-05,
+      "loss": 1.3525,
+      "step": 2214
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2468555120782061e-05,
+      "loss": 1.1526,
+      "step": 2215
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2462048115647209e-05,
+      "loss": 1.3355,
+      "step": 2216
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.24555400005962e-05,
+      "loss": 1.1323,
+      "step": 2217
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2449030778562954e-05,
+      "loss": 1.1422,
+      "step": 2218
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.244252045248189e-05,
+      "loss": 1.2039,
+      "step": 2219
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2436009025287939e-05,
+      "loss": 1.1257,
+      "step": 2220
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2429496499916512e-05,
+      "loss": 1.1653,
+      "step": 2221
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.242298287930352e-05,
+      "loss": 1.3818,
+      "step": 2222
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.241646816638537e-05,
+      "loss": 1.157,
+      "step": 2223
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2409952364098962e-05,
+      "loss": 1.3501,
+      "step": 2224
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2403435475381685e-05,
+      "loss": 1.2849,
+      "step": 2225
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.239691750317142e-05,
+      "loss": 1.3461,
+      "step": 2226
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.239039845040653e-05,
+      "loss": 1.2765,
+      "step": 2227
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2383878320025873e-05,
+      "loss": 0.9825,
+      "step": 2228
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2377357114968788e-05,
+      "loss": 1.2014,
+      "step": 2229
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2370834838175105e-05,
+      "loss": 1.1737,
+      "step": 2230
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2364311492585124e-05,
+      "loss": 1.2243,
+      "step": 2231
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2357787081139642e-05,
+      "loss": 1.2027,
+      "step": 2232
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2351261606779927e-05,
+      "loss": 1.3327,
+      "step": 2233
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2344735072447729e-05,
+      "loss": 1.0947,
+      "step": 2234
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2338207481085274e-05,
+      "loss": 1.2912,
+      "step": 2235
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2331678835635271e-05,
+      "loss": 1.1157,
+      "step": 2236
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2325149139040894e-05,
+      "loss": 1.342,
+      "step": 2237
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2318618394245798e-05,
+      "loss": 1.2068,
+      "step": 2238
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2312086604194107e-05,
+      "loss": 1.2986,
+      "step": 2239
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2305553771830422e-05,
+      "loss": 1.1082,
+      "step": 2240
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.229901990009981e-05,
+      "loss": 1.0305,
+      "step": 2241
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2292484991947802e-05,
+      "loss": 1.1277,
+      "step": 2242
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.22859490503204e-05,
+      "loss": 1.0156,
+      "step": 2243
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.227941207816408e-05,
+      "loss": 1.2787,
+      "step": 2244
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.2272874078425765e-05,
+      "loss": 1.1142,
+      "step": 2245
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.2266335054052857e-05,
+      "loss": 1.0921,
+      "step": 2246
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.2259795007993218e-05,
+      "loss": 1.2939,
+      "step": 2247
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.2253253943195161e-05,
+      "loss": 1.1929,
+      "step": 2248
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.2246711862607465e-05,
+      "loss": 1.5293,
+      "step": 2249
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.2240168769179365e-05,
+      "loss": 1.1558,
+      "step": 2250
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.2233624665860557e-05,
+      "loss": 1.4066,
+      "step": 2251
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.2227079555601188e-05,
+      "loss": 1.316,
+      "step": 2252
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.222053344135186e-05,
+      "loss": 1.2775,
+      "step": 2253
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.2213986326063622e-05,
+      "loss": 0.9638,
+      "step": 2254
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.2207438212687988e-05,
+      "loss": 1.2904,
+      "step": 2255
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.220088910417691e-05,
+      "loss": 1.1497,
+      "step": 2256
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.219433900348279e-05,
+      "loss": 1.3698,
+      "step": 2257
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.218778791355848e-05,
+      "loss": 0.993,
+      "step": 2258
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.2181235837357282e-05,
+      "loss": 1.3483,
+      "step": 2259
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.2174682777832935e-05,
+      "loss": 1.1403,
+      "step": 2260
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.2168128737939625e-05,
+      "loss": 1.1443,
+      "step": 2261
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.2161573720631976e-05,
+      "loss": 1.158,
+      "step": 2262
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.2155017728865064e-05,
+      "loss": 1.1663,
+      "step": 2263
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.2148460765594388e-05,
+      "loss": 1.1011,
+      "step": 2264
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.2141902833775898e-05,
+      "loss": 1.4067,
+      "step": 2265
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.2135343936365975e-05,
+      "loss": 1.0096,
+      "step": 2266
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.2128784076321434e-05,
+      "loss": 1.2742,
+      "step": 2267
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.212222325659953e-05,
+      "loss": 1.386,
+      "step": 2268
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.2115661480157944e-05,
+      "loss": 0.9602,
+      "step": 2269
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.2109098749954791e-05,
+      "loss": 1.2134,
+      "step": 2270
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.2102535068948618e-05,
+      "loss": 1.211,
+      "step": 2271
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.2095970440098398e-05,
+      "loss": 1.1523,
+      "step": 2272
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.208940486636353e-05,
+      "loss": 1.1991,
+      "step": 2273
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.2082838350703842e-05,
+      "loss": 1.1172,
+      "step": 2274
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.207627089607959e-05,
+      "loss": 1.2325,
+      "step": 2275
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.2069702505451444e-05,
+      "loss": 1.158,
+      "step": 2276
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.2063133181780499e-05,
+      "loss": 1.276,
+      "step": 2277
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.2056562928028277e-05,
+      "loss": 1.1128,
+      "step": 2278
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.2049991747156713e-05,
+      "loss": 1.1842,
+      "step": 2279
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.2043419642128162e-05,
+      "loss": 1.3659,
+      "step": 2280
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.2036846615905395e-05,
+      "loss": 1.0598,
+      "step": 2281
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.20302726714516e-05,
+      "loss": 1.0916,
+      "step": 2282
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.2023697811730372e-05,
+      "loss": 1.1247,
+      "step": 2283
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.2017122039705736e-05,
+      "loss": 1.0364,
+      "step": 2284
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.2010545358342102e-05,
+      "loss": 1.2245,
+      "step": 2285
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.2003967770604316e-05,
+      "loss": 1.1979,
+      "step": 2286
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1997389279457614e-05,
+      "loss": 1.2471,
+      "step": 2287
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1990809887867652e-05,
+      "loss": 1.0625,
+      "step": 2288
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1984229598800478e-05,
+      "loss": 1.1604,
+      "step": 2289
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.197764841522256e-05,
+      "loss": 1.2582,
+      "step": 2290
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.197106634010076e-05,
+      "loss": 1.1824,
+      "step": 2291
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1964483376402345e-05,
+      "loss": 1.1512,
+      "step": 2292
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1957899527094976e-05,
+      "loss": 1.1015,
+      "step": 2293
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1951314795146725e-05,
+      "loss": 1.0325,
+      "step": 2294
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1944729183526056e-05,
+      "loss": 1.1883,
+      "step": 2295
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1938142695201824e-05,
+      "loss": 1.0215,
+      "step": 2296
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1931555333143286e-05,
+      "loss": 1.2755,
+      "step": 2297
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1924967100320095e-05,
+      "loss": 1.2593,
+      "step": 2298
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1918377999702292e-05,
+      "loss": 1.1779,
+      "step": 2299
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1911788034260305e-05,
+      "loss": 1.2935,
+      "step": 2300
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1905197206964963e-05,
+      "loss": 1.1974,
+      "step": 2301
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1898605520787476e-05,
+      "loss": 1.1175,
+      "step": 2302
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.189201297869944e-05,
+      "loss": 1.2702,
+      "step": 2303
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1885419583672843e-05,
+      "loss": 1.2793,
+      "step": 2304
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1878825338680051e-05,
+      "loss": 1.1342,
+      "step": 2305
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1872230246693816e-05,
+      "loss": 1.3302,
+      "step": 2306
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1865634310687279e-05,
+      "loss": 1.1319,
+      "step": 2307
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1859037533633947e-05,
+      "loss": 1.4878,
+      "step": 2308
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1852439918507714e-05,
+      "loss": 1.3085,
+      "step": 2309
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1845841468282858e-05,
+      "loss": 1.3807,
+      "step": 2310
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1839242185934022e-05,
+      "loss": 1.1723,
+      "step": 2311
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.183264207443623e-05,
+      "loss": 1.1693,
+      "step": 2312
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1826041136764873e-05,
+      "loss": 1.274,
+      "step": 2313
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1819439375895731e-05,
+      "loss": 1.0779,
+      "step": 2314
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.181283679480494e-05,
+      "loss": 1.2659,
+      "step": 2315
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1806233396469008e-05,
+      "loss": 1.2445,
+      "step": 2316
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1799629183864816e-05,
+      "loss": 1.2261,
+      "step": 2317
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1793024159969614e-05,
+      "loss": 1.3361,
+      "step": 2318
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1786418327761002e-05,
+      "loss": 1.1765,
+      "step": 2319
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1779811690216968e-05,
+      "loss": 1.188,
+      "step": 2320
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1773204250315841e-05,
+      "loss": 1.0975,
+      "step": 2321
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1766596011036329e-05,
+      "loss": 1.1265,
+      "step": 2322
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1759986975357491e-05,
+      "loss": 1.4224,
+      "step": 2323
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1753377146258745e-05,
+      "loss": 1.1429,
+      "step": 2324
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1746766526719869e-05,
+      "loss": 1.1395,
+      "step": 2325
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1740155119721002e-05,
+      "loss": 1.2484,
+      "step": 2326
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1733542928242628e-05,
+      "loss": 1.2587,
+      "step": 2327
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1726929955265586e-05,
+      "loss": 1.2072,
+      "step": 2328
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1720316203771081e-05,
+      "loss": 1.2921,
+      "step": 2329
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1713701676740655e-05,
+      "loss": 1.0466,
+      "step": 2330
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.17070863771562e-05,
+      "loss": 1.0085,
+      "step": 2331
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1700470307999962e-05,
+      "loss": 1.4527,
+      "step": 2332
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1693853472254532e-05,
+      "loss": 1.1484,
+      "step": 2333
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1687235872902851e-05,
+      "loss": 1.2925,
+      "step": 2334
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.168061751292819e-05,
+      "loss": 1.1937,
+      "step": 2335
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.167399839531418e-05,
+      "loss": 1.2423,
+      "step": 2336
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1667378523044786e-05,
+      "loss": 0.9833,
+      "step": 2337
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1660757899104307e-05,
+      "loss": 1.2029,
+      "step": 2338
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1654136526477395e-05,
+      "loss": 1.1102,
+      "step": 2339
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1647514408149026e-05,
+      "loss": 1.2791,
+      "step": 2340
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1640891547104521e-05,
+      "loss": 1.1783,
+      "step": 2341
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.163426794632953e-05,
+      "loss": 1.343,
+      "step": 2342
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1627643608810041e-05,
+      "loss": 1.3292,
+      "step": 2343
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1621018537532372e-05,
+      "loss": 1.0665,
+      "step": 2344
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.161439273548317e-05,
+      "loss": 1.1005,
+      "step": 2345
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.160776620564942e-05,
+      "loss": 1.1117,
+      "step": 2346
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1601138951018422e-05,
+      "loss": 1.3396,
+      "step": 2347
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1594510974577808e-05,
+      "loss": 1.0432,
+      "step": 2348
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1587882279315544e-05,
+      "loss": 1.3237,
+      "step": 2349
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1581252868219908e-05,
+      "loss": 1.1273,
+      "step": 2350
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1574622744279507e-05,
+      "loss": 1.0497,
+      "step": 2351
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1567991910483264e-05,
+      "loss": 1.0323,
+      "step": 2352
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1561360369820434e-05,
+      "loss": 1.1204,
+      "step": 2353
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1554728125280574e-05,
+      "loss": 1.1148,
+      "step": 2354
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1548095179853572e-05,
+      "loss": 1.1649,
+      "step": 2355
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1541461536529623e-05,
+      "loss": 1.1886,
+      "step": 2356
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1534827198299245e-05,
+      "loss": 1.061,
+      "step": 2357
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.152819216815326e-05,
+      "loss": 1.3475,
+      "step": 2358
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.152155644908281e-05,
+      "loss": 1.3779,
+      "step": 2359
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1514920044079346e-05,
+      "loss": 1.0894,
+      "step": 2360
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.150828295613462e-05,
+      "loss": 1.1736,
+      "step": 2361
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1501645188240702e-05,
+      "loss": 1.0388,
+      "step": 2362
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1495006743389968e-05,
+      "loss": 1.0636,
+      "step": 2363
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1488367624575085e-05,
+      "loss": 1.2403,
+      "step": 2364
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.148172783478905e-05,
+      "loss": 1.1099,
+      "step": 2365
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1475087377025138e-05,
+      "loss": 1.2692,
+      "step": 2366
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1468446254276935e-05,
+      "loss": 1.1552,
+      "step": 2367
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1461804469538327e-05,
+      "loss": 1.1266,
+      "step": 2368
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.14551620258035e-05,
+      "loss": 1.1086,
+      "step": 2369
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1448518926066932e-05,
+      "loss": 1.1414,
+      "step": 2370
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1441875173323398e-05,
+      "loss": 1.2518,
+      "step": 2371
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1435230770567974e-05,
+      "loss": 1.1813,
+      "step": 2372
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1428585720796019e-05,
+      "loss": 0.9426,
+      "step": 2373
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1421940027003191e-05,
+      "loss": 1.2192,
+      "step": 2374
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1415293692185438e-05,
+      "loss": 1.2249,
+      "step": 2375
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1408646719338989e-05,
+      "loss": 1.0822,
+      "step": 2376
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1401999111460368e-05,
+      "loss": 1.2439,
+      "step": 2377
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1395350871546392e-05,
+      "loss": 1.1972,
+      "step": 2378
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1388702002594144e-05,
+      "loss": 1.2153,
+      "step": 2379
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1382052507601005e-05,
+      "loss": 1.1958,
+      "step": 2380
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1375402389564639e-05,
+      "loss": 1.1858,
+      "step": 2381
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.136875165148298e-05,
+      "loss": 1.2279,
+      "step": 2382
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.136210029635425e-05,
+      "loss": 1.1303,
+      "step": 2383
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1355448327176946e-05,
+      "loss": 1.1581,
+      "step": 2384
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1348795746949848e-05,
+      "loss": 1.0556,
+      "step": 2385
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1342142558672003e-05,
+      "loss": 1.3331,
+      "step": 2386
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.133548876534273e-05,
+      "loss": 1.2363,
+      "step": 2387
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1328834369961635e-05,
+      "loss": 0.9403,
+      "step": 2388
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1322179375528583e-05,
+      "loss": 1.0787,
+      "step": 2389
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.1315523785043712e-05,
+      "loss": 1.1523,
+      "step": 2390
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.130886760150743e-05,
+      "loss": 1.2071,
+      "step": 2391
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.1302210827920415e-05,
+      "loss": 1.0392,
+      "step": 2392
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.12955534672836e-05,
+      "loss": 1.0685,
+      "step": 2393
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.1288895522598198e-05,
+      "loss": 1.3955,
+      "step": 2394
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.1282236996865675e-05,
+      "loss": 1.5226,
+      "step": 2395
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.127557789308776e-05,
+      "loss": 1.0224,
+      "step": 2396
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.126891821426645e-05,
+      "loss": 1.1849,
+      "step": 2397
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.1262257963403988e-05,
+      "loss": 1.1739,
+      "step": 2398
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.1255597143502885e-05,
+      "loss": 1.1801,
+      "step": 2399
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.1248935757565906e-05,
+      "loss": 1.2495,
+      "step": 2400
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.1242273808596076e-05,
+      "loss": 1.075,
+      "step": 2401
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.1235611299596663e-05,
+      "loss": 1.214,
+      "step": 2402
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.122894823357119e-05,
+      "loss": 1.0312,
+      "step": 2403
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.1222284613523444e-05,
+      "loss": 1.2445,
+      "step": 2404
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.121562044245745e-05,
+      "loss": 1.2094,
+      "step": 2405
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.1208955723377474e-05,
+      "loss": 1.2333,
+      "step": 2406
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.120229045928805e-05,
+      "loss": 1.4344,
+      "step": 2407
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.1195624653193944e-05,
+      "loss": 1.0895,
+      "step": 2408
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.1188958308100163e-05,
+      "loss": 1.0559,
+      "step": 2409
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.118229142701197e-05,
+      "loss": 1.2226,
+      "step": 2410
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.1175624012934858e-05,
+      "loss": 1.1255,
+      "step": 2411
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.1168956068874569e-05,
+      "loss": 1.1292,
+      "step": 2412
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.116228759783707e-05,
+      "loss": 1.1375,
+      "step": 2413
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.1155618602828586e-05,
+      "loss": 1.039,
+      "step": 2414
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.1148949086855561e-05,
+      "loss": 1.1499,
+      "step": 2415
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.1142279052924678e-05,
+      "loss": 1.1266,
+      "step": 2416
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.1135608504042865e-05,
+      "loss": 1.1253,
+      "step": 2417
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.1128937443217265e-05,
+      "loss": 1.1019,
+      "step": 2418
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.1122265873455262e-05,
+      "loss": 1.4146,
+      "step": 2419
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.1115593797764468e-05,
+      "loss": 1.0064,
+      "step": 2420
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.1108921219152721e-05,
+      "loss": 1.1916,
+      "step": 2421
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.1102248140628084e-05,
+      "loss": 1.1192,
+      "step": 2422
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.1095574565198855e-05,
+      "loss": 1.138,
+      "step": 2423
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.1088900495873545e-05,
+      "loss": 1.1639,
+      "step": 2424
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.108222593566089e-05,
+      "loss": 1.2299,
+      "step": 2425
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.1075550887569853e-05,
+      "loss": 1.1605,
+      "step": 2426
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.1068875354609608e-05,
+      "loss": 1.0904,
+      "step": 2427
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.1062199339789562e-05,
+      "loss": 1.2084,
+      "step": 2428
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.1055522846119317e-05,
+      "loss": 0.9919,
+      "step": 2429
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.1048845876608713e-05,
+      "loss": 1.1639,
+      "step": 2430
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.1042168434267797e-05,
+      "loss": 1.3598,
+      "step": 2431
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.1035490522106822e-05,
+      "loss": 1.2158,
+      "step": 2432
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.102881214313626e-05,
+      "loss": 1.1939,
+      "step": 2433
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.1022133300366788e-05,
+      "loss": 1.0961,
+      "step": 2434
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.1015453996809302e-05,
+      "loss": 1.0663,
+      "step": 2435
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.10087742354749e-05,
+      "loss": 1.0006,
+      "step": 2436
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.100209401937488e-05,
+      "loss": 1.3639,
+      "step": 2437
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0995413351520752e-05,
+      "loss": 1.0063,
+      "step": 2438
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0988732234924232e-05,
+      "loss": 1.2911,
+      "step": 2439
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0982050672597233e-05,
+      "loss": 1.3328,
+      "step": 2440
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.097536866755187e-05,
+      "loss": 1.2163,
+      "step": 2441
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0968686222800457e-05,
+      "loss": 1.0909,
+      "step": 2442
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0962003341355514e-05,
+      "loss": 1.2299,
+      "step": 2443
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0955320026229747e-05,
+      "loss": 1.1132,
+      "step": 2444
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.094863628043606e-05,
+      "loss": 1.3241,
+      "step": 2445
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0941952106987558e-05,
+      "loss": 1.1746,
+      "step": 2446
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0935267508897528e-05,
+      "loss": 0.9526,
+      "step": 2447
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.092858248917946e-05,
+      "loss": 1.2502,
+      "step": 2448
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0921897050847025e-05,
+      "loss": 1.1272,
+      "step": 2449
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0915211196914086e-05,
+      "loss": 1.176,
+      "step": 2450
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0908524930394696e-05,
+      "loss": 1.2127,
+      "step": 2451
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0901838254303093e-05,
+      "loss": 1.2468,
+      "step": 2452
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0895151171653693e-05,
+      "loss": 1.3976,
+      "step": 2453
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0888463685461099e-05,
+      "loss": 1.1882,
+      "step": 2454
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0881775798740104e-05,
+      "loss": 1.1823,
+      "step": 2455
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0875087514505674e-05,
+      "loss": 1.2254,
+      "step": 2456
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0868398835772949e-05,
+      "loss": 1.2134,
+      "step": 2457
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0861709765557257e-05,
+      "loss": 1.0532,
+      "step": 2458
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0855020306874102e-05,
+      "loss": 1.1066,
+      "step": 2459
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0848330462739155e-05,
+      "loss": 1.2646,
+      "step": 2460
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0841640236168265e-05,
+      "loss": 1.1653,
+      "step": 2461
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0834949630177453e-05,
+      "loss": 1.3713,
+      "step": 2462
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0828258647782917e-05,
+      "loss": 1.2307,
+      "step": 2463
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0821567292001013e-05,
+      "loss": 1.2636,
+      "step": 2464
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.081487556584828e-05,
+      "loss": 1.2957,
+      "step": 2465
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0808183472341405e-05,
+      "loss": 1.1887,
+      "step": 2466
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.080149101449726e-05,
+      "loss": 0.9169,
+      "step": 2467
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.079479819533287e-05,
+      "loss": 1.1588,
+      "step": 2468
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0788105017865422e-05,
+      "loss": 1.2526,
+      "step": 2469
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0781411485112274e-05,
+      "loss": 1.1792,
+      "step": 2470
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0774717600090934e-05,
+      "loss": 1.0816,
+      "step": 2471
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0768023365819077e-05,
+      "loss": 1.1534,
+      "step": 2472
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0761328785314525e-05,
+      "loss": 1.0546,
+      "step": 2473
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0754633861595265e-05,
+      "loss": 1.1876,
+      "step": 2474
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0747938597679441e-05,
+      "loss": 0.9552,
+      "step": 2475
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0741242996585342e-05,
+      "loss": 1.5504,
+      "step": 2476
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.073454706133141e-05,
+      "loss": 1.1845,
+      "step": 2477
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0727850794936246e-05,
+      "loss": 1.1266,
+      "step": 2478
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0721154200418589e-05,
+      "loss": 1.1109,
+      "step": 2479
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0714457280797334e-05,
+      "loss": 1.1341,
+      "step": 2480
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0707760039091519e-05,
+      "loss": 1.204,
+      "step": 2481
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0701062478320331e-05,
+      "loss": 1.1547,
+      "step": 2482
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.06943646015031e-05,
+      "loss": 1.1606,
+      "step": 2483
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0687666411659288e-05,
+      "loss": 1.1699,
+      "step": 2484
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0680967911808509e-05,
+      "loss": 1.1529,
+      "step": 2485
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.067426910497052e-05,
+      "loss": 1.1002,
+      "step": 2486
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0667569994165207e-05,
+      "loss": 1.0629,
+      "step": 2487
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0660870582412598e-05,
+      "loss": 1.0988,
+      "step": 2488
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0654170872732854e-05,
+      "loss": 1.4567,
+      "step": 2489
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0647470868146272e-05,
+      "loss": 1.201,
+      "step": 2490
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0640770571673284e-05,
+      "loss": 1.1936,
+      "step": 2491
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.063406998633445e-05,
+      "loss": 0.9668,
+      "step": 2492
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.062736911515046e-05,
+      "loss": 1.1681,
+      "step": 2493
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0620667961142135e-05,
+      "loss": 1.0111,
+      "step": 2494
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0613966527330426e-05,
+      "loss": 1.5403,
+      "step": 2495
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0607264816736404e-05,
+      "loss": 1.3356,
+      "step": 2496
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0600562832381265e-05,
+      "loss": 1.2245,
+      "step": 2497
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.059386057728634e-05,
+      "loss": 1.2306,
+      "step": 2498
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0587158054473064e-05,
+      "loss": 1.0849,
+      "step": 2499
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0580455266963008e-05,
+      "loss": 1.2756,
+      "step": 2500
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.057375221777785e-05,
+      "loss": 1.075,
+      "step": 2501
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0567048909939403e-05,
+      "loss": 1.1643,
+      "step": 2502
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0560345346469577e-05,
+      "loss": 1.1526,
+      "step": 2503
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0553641530390409e-05,
+      "loss": 1.1936,
+      "step": 2504
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0546937464724042e-05,
+      "loss": 1.2528,
+      "step": 2505
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0540233152492747e-05,
+      "loss": 1.259,
+      "step": 2506
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0533528596718887e-05,
+      "loss": 0.9533,
+      "step": 2507
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.052682380042495e-05,
+      "loss": 1.4487,
+      "step": 2508
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0520118766633519e-05,
+      "loss": 1.175,
+      "step": 2509
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0513413498367293e-05,
+      "loss": 1.0196,
+      "step": 2510
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0506707998649076e-05,
+      "loss": 1.1072,
+      "step": 2511
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0500002270501777e-05,
+      "loss": 1.4025,
+      "step": 2512
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0493296316948403e-05,
+      "loss": 1.1908,
+      "step": 2513
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0486590141012065e-05,
+      "loss": 1.2393,
+      "step": 2514
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.047988374571598e-05,
+      "loss": 1.0564,
+      "step": 2515
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0473177134083452e-05,
+      "loss": 1.1725,
+      "step": 2516
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0466470309137891e-05,
+      "loss": 1.0899,
+      "step": 2517
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0459763273902806e-05,
+      "loss": 1.1981,
+      "step": 2518
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0453056031401791e-05,
+      "loss": 1.1831,
+      "step": 2519
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.044634858465854e-05,
+      "loss": 1.0864,
+      "step": 2520
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.043964093669684e-05,
+      "loss": 1.0047,
+      "step": 2521
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0432933090540563e-05,
+      "loss": 1.2436,
+      "step": 2522
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0426225049213678e-05,
+      "loss": 1.2942,
+      "step": 2523
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0419516815740234e-05,
+      "loss": 1.2555,
+      "step": 2524
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0412808393144374e-05,
+      "loss": 1.1223,
+      "step": 2525
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0406099784450317e-05,
+      "loss": 1.2168,
+      "step": 2526
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0399390992682377e-05,
+      "loss": 1.226,
+      "step": 2527
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0392682020864941e-05,
+      "loss": 1.2135,
+      "step": 2528
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0385972872022485e-05,
+      "loss": 1.3024,
+      "step": 2529
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.037926354917956e-05,
+      "loss": 1.0888,
+      "step": 2530
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0372554055360797e-05,
+      "loss": 1.146,
+      "step": 2531
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.03658443935909e-05,
+      "loss": 1.141,
+      "step": 2532
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0359134566894656e-05,
+      "loss": 1.0781,
+      "step": 2533
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0352424578296922e-05,
+      "loss": 1.4046,
+      "step": 2534
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0345714430822627e-05,
+      "loss": 1.257,
+      "step": 2535
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0339004127496775e-05,
+      "loss": 1.2857,
+      "step": 2536
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0332293671344438e-05,
+      "loss": 1.1538,
+      "step": 2537
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.0325583065390759e-05,
+      "loss": 1.164,
+      "step": 2538
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.0318872312660944e-05,
+      "loss": 1.0617,
+      "step": 2539
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.0312161416180266e-05,
+      "loss": 1.0614,
+      "step": 2540
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.0305450378974074e-05,
+      "loss": 1.0877,
+      "step": 2541
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.0298739204067762e-05,
+      "loss": 1.3998,
+      "step": 2542
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.0292027894486802e-05,
+      "loss": 1.2237,
+      "step": 2543
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.0285316453256715e-05,
+      "loss": 1.2438,
+      "step": 2544
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.027860488340309e-05,
+      "loss": 1.032,
+      "step": 2545
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.027189318795157e-05,
+      "loss": 1.1998,
+      "step": 2546
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.0265181369927853e-05,
+      "loss": 1.0007,
+      "step": 2547
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.0258469432357694e-05,
+      "loss": 1.2771,
+      "step": 2548
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.0251757378266906e-05,
+      "loss": 1.0073,
+      "step": 2549
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.024504521068135e-05,
+      "loss": 0.9386,
+      "step": 2550
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.0238332932626935e-05,
+      "loss": 1.1129,
+      "step": 2551
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.0231620547129628e-05,
+      "loss": 1.1362,
+      "step": 2552
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.0224908057215442e-05,
+      "loss": 1.0301,
+      "step": 2553
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.0218195465910428e-05,
+      "loss": 1.2322,
+      "step": 2554
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.0211482776240693e-05,
+      "loss": 1.1792,
+      "step": 2555
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.0204769991232388e-05,
+      "loss": 1.191,
+      "step": 2556
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.0198057113911703e-05,
+      "loss": 1.174,
+      "step": 2557
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.019134414730487e-05,
+      "loss": 1.1914,
+      "step": 2558
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.0184631094438162e-05,
+      "loss": 1.1397,
+      "step": 2559
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.0177917958337892e-05,
+      "loss": 1.0512,
+      "step": 2560
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.0171204742030409e-05,
+      "loss": 1.2283,
+      "step": 2561
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.01644914485421e-05,
+      "loss": 1.2756,
+      "step": 2562
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.0157778080899383e-05,
+      "loss": 1.0485,
+      "step": 2563
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.0151064642128711e-05,
+      "loss": 1.1418,
+      "step": 2564
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.0144351135256577e-05,
+      "loss": 0.9848,
+      "step": 2565
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.0137637563309494e-05,
+      "loss": 1.0734,
+      "step": 2566
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.0130923929314003e-05,
+      "loss": 1.3019,
+      "step": 2567
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.0124210236296682e-05,
+      "loss": 1.1405,
+      "step": 2568
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.0117496487284134e-05,
+      "loss": 1.1821,
+      "step": 2569
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.0110782685302984e-05,
+      "loss": 1.34,
+      "step": 2570
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.0104068833379878e-05,
+      "loss": 1.2881,
+      "step": 2571
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.0097354934541492e-05,
+      "loss": 0.9109,
+      "step": 2572
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.0090640991814518e-05,
+      "loss": 1.3788,
+      "step": 2573
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.0083927008225666e-05,
+      "loss": 1.2591,
+      "step": 2574
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.0077212986801674e-05,
+      "loss": 1.0945,
+      "step": 2575
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.0070498930569287e-05,
+      "loss": 1.2679,
+      "step": 2576
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.0063784842555272e-05,
+      "loss": 1.4073,
+      "step": 2577
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.0057070725786403e-05,
+      "loss": 1.301,
+      "step": 2578
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.0050356583289475e-05,
+      "loss": 1.0238,
+      "step": 2579
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.0043642418091293e-05,
+      "loss": 1.2873,
+      "step": 2580
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.0036928233218664e-05,
+      "loss": 1.021,
+      "step": 2581
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.0030214031698418e-05,
+      "loss": 1.337,
+      "step": 2582
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.0023499816557378e-05,
+      "loss": 1.1581,
+      "step": 2583
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.0016785590822385e-05,
+      "loss": 1.1967,
+      "step": 2584
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.001007135752028e-05,
+      "loss": 1.1492,
+      "step": 2585
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.0003357119677904e-05,
+      "loss": 1.2608,
+      "step": 2586
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.996642880322101e-06,
+      "loss": 1.029,
+      "step": 2587
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.989928642479724e-06,
+      "loss": 1.2388,
+      "step": 2588
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.983214409177617e-06,
+      "loss": 1.1279,
+      "step": 2589
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.976500183442625e-06,
+      "loss": 1.3491,
+      "step": 2590
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.969785968301585e-06,
+      "loss": 1.1686,
+      "step": 2591
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.963071766781338e-06,
+      "loss": 1.3053,
+      "step": 2592
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.956357581908713e-06,
+      "loss": 1.4022,
+      "step": 2593
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.949643416710526e-06,
+      "loss": 1.2659,
+      "step": 2594
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.942929274213598e-06,
+      "loss": 1.1317,
+      "step": 2595
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.936215157444732e-06,
+      "loss": 1.1397,
+      "step": 2596
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.929501069430714e-06,
+      "loss": 1.0968,
+      "step": 2597
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.922787013198325e-06,
+      "loss": 1.0523,
+      "step": 2598
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.916072991774335e-06,
+      "loss": 1.1084,
+      "step": 2599
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.909359008185486e-06,
+      "loss": 1.3072,
+      "step": 2600
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.902645065458514e-06,
+      "loss": 1.0771,
+      "step": 2601
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.895931166620125e-06,
+      "loss": 1.4577,
+      "step": 2602
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.889217314697018e-06,
+      "loss": 1.343,
+      "step": 2603
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.882503512715869e-06,
+      "loss": 0.962,
+      "step": 2604
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.87578976370332e-06,
+      "loss": 1.1605,
+      "step": 2605
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.869076070686e-06,
+      "loss": 1.3753,
+      "step": 2606
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.862362436690511e-06,
+      "loss": 1.242,
+      "step": 2607
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.855648864743424e-06,
+      "loss": 0.9864,
+      "step": 2608
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.848935357871289e-06,
+      "loss": 1.1393,
+      "step": 2609
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.84222191910062e-06,
+      "loss": 1.317,
+      "step": 2610
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.835508551457903e-06,
+      "loss": 1.3485,
+      "step": 2611
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.828795257969593e-06,
+      "loss": 1.1985,
+      "step": 2612
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.822082041662113e-06,
+      "loss": 1.2935,
+      "step": 2613
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.81536890556184e-06,
+      "loss": 1.3371,
+      "step": 2614
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.808655852695132e-06,
+      "loss": 1.1138,
+      "step": 2615
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.801942886088302e-06,
+      "loss": 1.2265,
+      "step": 2616
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.795230008767614e-06,
+      "loss": 1.2867,
+      "step": 2617
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.788517223759309e-06,
+      "loss": 1.1942,
+      "step": 2618
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.781804534089577e-06,
+      "loss": 1.2365,
+      "step": 2619
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.775091942784563e-06,
+      "loss": 1.2078,
+      "step": 2620
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.768379452870372e-06,
+      "loss": 1.5032,
+      "step": 2621
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.761667067373068e-06,
+      "loss": 1.3221,
+      "step": 2622
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.754954789318653e-06,
+      "loss": 1.0478,
+      "step": 2623
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.748242621733094e-06,
+      "loss": 1.3939,
+      "step": 2624
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.74153056764231e-06,
+      "loss": 1.3683,
+      "step": 2625
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.734818630072149e-06,
+      "loss": 1.209,
+      "step": 2626
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.728106812048432e-06,
+      "loss": 1.0883,
+      "step": 2627
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.721395116596914e-06,
+      "loss": 1.2743,
+      "step": 2628
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.714683546743288e-06,
+      "loss": 1.2651,
+      "step": 2629
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.707972105513203e-06,
+      "loss": 1.2423,
+      "step": 2630
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.701260795932241e-06,
+      "loss": 1.2396,
+      "step": 2631
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.694549621025928e-06,
+      "loss": 1.1089,
+      "step": 2632
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.687838583819735e-06,
+      "loss": 1.5084,
+      "step": 2633
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.681127687339061e-06,
+      "loss": 1.2494,
+      "step": 2634
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.674416934609243e-06,
+      "loss": 1.1506,
+      "step": 2635
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.667706328655567e-06,
+      "loss": 1.032,
+      "step": 2636
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.660995872503229e-06,
+      "loss": 1.031,
+      "step": 2637
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.654285569177373e-06,
+      "loss": 1.324,
+      "step": 2638
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.647575421703083e-06,
+      "loss": 1.1488,
+      "step": 2639
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.640865433105348e-06,
+      "loss": 1.3003,
+      "step": 2640
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.634155606409102e-06,
+      "loss": 1.176,
+      "step": 2641
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.627445944639208e-06,
+      "loss": 1.1483,
+      "step": 2642
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.620736450820443e-06,
+      "loss": 1.1784,
+      "step": 2643
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.614027127977515e-06,
+      "loss": 0.9816,
+      "step": 2644
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.60731797913506e-06,
+      "loss": 1.2676,
+      "step": 2645
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.600609007317627e-06,
+      "loss": 1.052,
+      "step": 2646
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.593900215549685e-06,
+      "loss": 1.2228,
+      "step": 2647
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.587191606855633e-06,
+      "loss": 1.2255,
+      "step": 2648
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.580483184259769e-06,
+      "loss": 1.1658,
+      "step": 2649
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.573774950786322e-06,
+      "loss": 1.355,
+      "step": 2650
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.567066909459439e-06,
+      "loss": 1.3526,
+      "step": 2651
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.560359063303163e-06,
+      "loss": 1.3349,
+      "step": 2652
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.553651415341461e-06,
+      "loss": 1.2074,
+      "step": 2653
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.546943968598212e-06,
+      "loss": 1.0441,
+      "step": 2654
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.540236726097196e-06,
+      "loss": 1.0582,
+      "step": 2655
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.533529690862109e-06,
+      "loss": 1.2252,
+      "step": 2656
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.526822865916553e-06,
+      "loss": 1.3074,
+      "step": 2657
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.520116254284023e-06,
+      "loss": 1.2325,
+      "step": 2658
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.513409858987939e-06,
+      "loss": 1.3514,
+      "step": 2659
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.506703683051602e-06,
+      "loss": 0.9447,
+      "step": 2660
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.499997729498226e-06,
+      "loss": 1.3313,
+      "step": 2661
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.493292001350927e-06,
+      "loss": 1.4135,
+      "step": 2662
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.486586501632712e-06,
+      "loss": 1.2326,
+      "step": 2663
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.479881233366485e-06,
+      "loss": 1.3627,
+      "step": 2664
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.473176199575056e-06,
+      "loss": 1.4012,
+      "step": 2665
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.466471403281115e-06,
+      "loss": 1.3291,
+      "step": 2666
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.459766847507254e-06,
+      "loss": 1.0681,
+      "step": 2667
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.453062535275961e-06,
+      "loss": 1.1725,
+      "step": 2668
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.446358469609594e-06,
+      "loss": 1.198,
+      "step": 2669
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.439654653530425e-06,
+      "loss": 1.0938,
+      "step": 2670
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.432951090060602e-06,
+      "loss": 1.2069,
+      "step": 2671
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.426247782222152e-06,
+      "loss": 1.3039,
+      "step": 2672
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.419544733036996e-06,
+      "loss": 1.2391,
+      "step": 2673
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.41284194552694e-06,
+      "loss": 1.1729,
+      "step": 2674
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.406139422713666e-06,
+      "loss": 1.438,
+      "step": 2675
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.399437167618736e-06,
+      "loss": 1.0462,
+      "step": 2676
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.392735183263601e-06,
+      "loss": 1.1946,
+      "step": 2677
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.386033472669577e-06,
+      "loss": 1.1617,
+      "step": 2678
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.379332038857866e-06,
+      "loss": 1.1408,
+      "step": 2679
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.372630884849546e-06,
+      "loss": 1.2507,
+      "step": 2680
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.365930013665552e-06,
+      "loss": 1.1627,
+      "step": 2681
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.359229428326716e-06,
+      "loss": 1.1952,
+      "step": 2682
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.352529131853731e-06,
+      "loss": 1.1528,
+      "step": 2683
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.345829127267148e-06,
+      "loss": 1.1679,
+      "step": 2684
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.339129417587403e-06,
+      "loss": 1.216,
+      "step": 2685
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.332430005834794e-06,
+      "loss": 1.1398,
+      "step": 2686
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.325730895029481e-06,
+      "loss": 1.2669,
+      "step": 2687
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.319032088191493e-06,
+      "loss": 1.008,
+      "step": 2688
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.312333588340716e-06,
+      "loss": 1.2546,
+      "step": 2689
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.305635398496904e-06,
+      "loss": 1.1145,
+      "step": 2690
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.298937521679672e-06,
+      "loss": 1.2855,
+      "step": 2691
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.292239960908482e-06,
+      "loss": 0.953,
+      "step": 2692
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.285542719202666e-06,
+      "loss": 1.2446,
+      "step": 2693
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.278845799581416e-06,
+      "loss": 1.0055,
+      "step": 2694
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.272149205063759e-06,
+      "loss": 1.2756,
+      "step": 2695
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.265452938668592e-06,
+      "loss": 1.2268,
+      "step": 2696
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.258757003414662e-06,
+      "loss": 1.1888,
+      "step": 2697
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.25206140232056e-06,
+      "loss": 1.1328,
+      "step": 2698
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.245366138404736e-06,
+      "loss": 1.1233,
+      "step": 2699
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.23867121468548e-06,
+      "loss": 1.0241,
+      "step": 2700
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.231976634180928e-06,
+      "loss": 0.9353,
+      "step": 2701
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.225282399909066e-06,
+      "loss": 1.0422,
+      "step": 2702
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.218588514887731e-06,
+      "loss": 0.9758,
+      "step": 2703
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.211894982134581e-06,
+      "loss": 1.4359,
+      "step": 2704
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.205201804667131e-06,
+      "loss": 0.9279,
+      "step": 2705
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.198508985502745e-06,
+      "loss": 1.4646,
+      "step": 2706
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.191816527658596e-06,
+      "loss": 1.245,
+      "step": 2707
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.185124434151725e-06,
+      "loss": 1.057,
+      "step": 2708
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.178432707998988e-06,
+      "loss": 1.1535,
+      "step": 2709
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.171741352217085e-06,
+      "loss": 1.3082,
+      "step": 2710
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.165050369822548e-06,
+      "loss": 1.0001,
+      "step": 2711
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.158359763831738e-06,
+      "loss": 1.0653,
+      "step": 2712
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.151669537260848e-06,
+      "loss": 0.968,
+      "step": 2713
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.144979693125903e-06,
+      "loss": 1.196,
+      "step": 2714
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.138290234442744e-06,
+      "loss": 0.9445,
+      "step": 2715
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.131601164227054e-06,
+      "loss": 0.9965,
+      "step": 2716
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.124912485494331e-06,
+      "loss": 1.2488,
+      "step": 2717
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.118224201259898e-06,
+      "loss": 1.2171,
+      "step": 2718
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.111536314538903e-06,
+      "loss": 1.111,
+      "step": 2719
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.104848828346314e-06,
+      "loss": 1.0289,
+      "step": 2720
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.098161745696912e-06,
+      "loss": 0.9361,
+      "step": 2721
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.091475069605304e-06,
+      "loss": 1.1998,
+      "step": 2722
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.084788803085915e-06,
+      "loss": 1.2642,
+      "step": 2723
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.078102949152977e-06,
+      "loss": 1.1228,
+      "step": 2724
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.071417510820541e-06,
+      "loss": 1.1363,
+      "step": 2725
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.064732491102477e-06,
+      "loss": 1.0675,
+      "step": 2726
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.058047893012447e-06,
+      "loss": 1.1165,
+      "step": 2727
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.051363719563943e-06,
+      "loss": 1.0452,
+      "step": 2728
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.044679973770258e-06,
+      "loss": 1.08,
+      "step": 2729
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.037996658644488e-06,
+      "loss": 1.3377,
+      "step": 2730
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.031313777199543e-06,
+      "loss": 1.1954,
+      "step": 2731
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.024631332448134e-06,
+      "loss": 1.1235,
+      "step": 2732
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.017949327402769e-06,
+      "loss": 0.9525,
+      "step": 2733
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.01126776507577e-06,
+      "loss": 1.1378,
+      "step": 2734
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.004586648479251e-06,
+      "loss": 1.1117,
+      "step": 2735
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.997905980625123e-06,
+      "loss": 1.1436,
+      "step": 2736
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.991225764525102e-06,
+      "loss": 1.0697,
+      "step": 2737
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.984546003190701e-06,
+      "loss": 1.3295,
+      "step": 2738
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.977866699633214e-06,
+      "loss": 1.2481,
+      "step": 2739
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.971187856863743e-06,
+      "loss": 1.0597,
+      "step": 2740
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.964509477893182e-06,
+      "loss": 1.2223,
+      "step": 2741
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.957831565732205e-06,
+      "loss": 1.1636,
+      "step": 2742
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.951154123391288e-06,
+      "loss": 1.1556,
+      "step": 2743
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.944477153880685e-06,
+      "loss": 1.1416,
+      "step": 2744
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.937800660210443e-06,
+      "loss": 0.9965,
+      "step": 2745
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.931124645390397e-06,
+      "loss": 1.3376,
+      "step": 2746
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.924449112430152e-06,
+      "loss": 1.2475,
+      "step": 2747
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.917774064339112e-06,
+      "loss": 1.2211,
+      "step": 2748
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.911099504126462e-06,
+      "loss": 1.2277,
+      "step": 2749
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.904425434801149e-06,
+      "loss": 1.349,
+      "step": 2750
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.897751859371917e-06,
+      "loss": 1.3307,
+      "step": 2751
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.891078780847284e-06,
+      "loss": 1.2484,
+      "step": 2752
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.884406202235536e-06,
+      "loss": 1.1243,
+      "step": 2753
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.87773412654474e-06,
+      "loss": 0.9725,
+      "step": 2754
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.871062556782739e-06,
+      "loss": 1.1278,
+      "step": 2755
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.864391495957138e-06,
+      "loss": 1.2936,
+      "step": 2756
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.85772094707532e-06,
+      "loss": 1.2637,
+      "step": 2757
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.851050913144445e-06,
+      "loss": 1.3374,
+      "step": 2758
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.844381397171417e-06,
+      "loss": 1.2644,
+      "step": 2759
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.83771240216293e-06,
+      "loss": 1.4646,
+      "step": 2760
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.831043931125438e-06,
+      "loss": 1.1648,
+      "step": 2761
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.824375987065145e-06,
+      "loss": 1.1005,
+      "step": 2762
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.817708572988031e-06,
+      "loss": 1.1796,
+      "step": 2763
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.811041691899838e-06,
+      "loss": 1.098,
+      "step": 2764
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.80437534680606e-06,
+      "loss": 1.3762,
+      "step": 2765
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.797709540711949e-06,
+      "loss": 1.1136,
+      "step": 2766
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.791044276622527e-06,
+      "loss": 1.1675,
+      "step": 2767
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.784379557542554e-06,
+      "loss": 1.0254,
+      "step": 2768
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.777715386476556e-06,
+      "loss": 1.1177,
+      "step": 2769
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.771051766428813e-06,
+      "loss": 1.0263,
+      "step": 2770
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.76438870040334e-06,
+      "loss": 1.0609,
+      "step": 2771
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.757726191403929e-06,
+      "loss": 1.2195,
+      "step": 2772
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.751064242434096e-06,
+      "loss": 1.2996,
+      "step": 2773
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.744402856497117e-06,
+      "loss": 1.0608,
+      "step": 2774
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.737742036596017e-06,
+      "loss": 1.2607,
+      "step": 2775
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.731081785733556e-06,
+      "loss": 1.2184,
+      "step": 2776
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.72442210691224e-06,
+      "loss": 1.36,
+      "step": 2777
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.717763003134329e-06,
+      "loss": 1.273,
+      "step": 2778
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.711104477401804e-06,
+      "loss": 1.0731,
+      "step": 2779
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.7044465327164e-06,
+      "loss": 1.0347,
+      "step": 2780
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.697789172079592e-06,
+      "loss": 1.1859,
+      "step": 2781
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.691132398492574e-06,
+      "loss": 1.1004,
+      "step": 2782
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.684476214956288e-06,
+      "loss": 1.1927,
+      "step": 2783
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.677820624471422e-06,
+      "loss": 1.1262,
+      "step": 2784
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.671165630038368e-06,
+      "loss": 1.1664,
+      "step": 2785
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.664511234657273e-06,
+      "loss": 1.244,
+      "step": 2786
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.657857441328004e-06,
+      "loss": 1.0724,
+      "step": 2787
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.651204253050157e-06,
+      "loss": 1.1207,
+      "step": 2788
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.644551672823055e-06,
+      "loss": 1.2635,
+      "step": 2789
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.637899703645753e-06,
+      "loss": 1.1783,
+      "step": 2790
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.631248348517023e-06,
+      "loss": 1.0615,
+      "step": 2791
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.624597610435363e-06,
+      "loss": 1.2574,
+      "step": 2792
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.617947492398998e-06,
+      "loss": 1.2881,
+      "step": 2793
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.61129799740586e-06,
+      "loss": 1.1493,
+      "step": 2794
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.60464912845361e-06,
+      "loss": 1.4515,
+      "step": 2795
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.598000888539633e-06,
+      "loss": 1.2586,
+      "step": 2796
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.591353280661015e-06,
+      "loss": 1.1651,
+      "step": 2797
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.584706307814567e-06,
+      "loss": 1.1319,
+      "step": 2798
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.578059972996812e-06,
+      "loss": 1.0394,
+      "step": 2799
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.571414279203983e-06,
+      "loss": 1.1833,
+      "step": 2800
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.564769229432033e-06,
+      "loss": 1.34,
+      "step": 2801
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.558124826676603e-06,
+      "loss": 1.058,
+      "step": 2802
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.551481073933072e-06,
+      "loss": 1.1745,
+      "step": 2803
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.544837974196506e-06,
+      "loss": 1.0143,
+      "step": 2804
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.538195530461677e-06,
+      "loss": 1.0937,
+      "step": 2805
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.53155374572307e-06,
+      "loss": 1.2224,
+      "step": 2806
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.524912622974867e-06,
+      "loss": 1.1934,
+      "step": 2807
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.518272165210954e-06,
+      "loss": 1.2035,
+      "step": 2808
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.511632375424913e-06,
+      "loss": 1.2026,
+      "step": 2809
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.504993256610037e-06,
+      "loss": 1.1577,
+      "step": 2810
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.4983548117593e-06,
+      "loss": 1.3399,
+      "step": 2811
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.491717043865382e-06,
+      "loss": 1.1756,
+      "step": 2812
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.485079955920661e-06,
+      "loss": 1.1481,
+      "step": 2813
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.47844355091719e-06,
+      "loss": 1.0921,
+      "step": 2814
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.47180783184674e-06,
+      "loss": 1.2252,
+      "step": 2815
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.46517280170076e-06,
+      "loss": 1.1223,
+      "step": 2816
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.45853846347038e-06,
+      "loss": 1.3217,
+      "step": 2817
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.451904820146431e-06,
+      "loss": 1.203,
+      "step": 2818
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.44527187471943e-06,
+      "loss": 1.2274,
+      "step": 2819
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.438639630179571e-06,
+      "loss": 1.1614,
+      "step": 2820
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.432008089516737e-06,
+      "loss": 1.1613,
+      "step": 2821
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.425377255720496e-06,
+      "loss": 1.3562,
+      "step": 2822
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.418747131780093e-06,
+      "loss": 1.1889,
+      "step": 2823
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.412117720684456e-06,
+      "loss": 1.3079,
+      "step": 2824
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.405489025422196e-06,
+      "loss": 1.0908,
+      "step": 2825
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.398861048981581e-06,
+      "loss": 1.2698,
+      "step": 2826
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.392233794350586e-06,
+      "loss": 1.0885,
+      "step": 2827
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.385607264516833e-06,
+      "loss": 1.2889,
+      "step": 2828
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.378981462467631e-06,
+      "loss": 1.2644,
+      "step": 2829
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.372356391189964e-06,
+      "loss": 1.2248,
+      "step": 2830
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.365732053670475e-06,
+      "loss": 1.2175,
+      "step": 2831
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.359108452895482e-06,
+      "loss": 1.1181,
+      "step": 2832
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.352485591850978e-06,
+      "loss": 1.2012,
+      "step": 2833
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.345863473522608e-06,
+      "loss": 1.2583,
+      "step": 2834
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.339242100895693e-06,
+      "loss": 1.189,
+      "step": 2835
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.332621476955221e-06,
+      "loss": 1.1651,
+      "step": 2836
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.326001604685822e-06,
+      "loss": 1.132,
+      "step": 2837
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.31938248707181e-06,
+      "loss": 1.129,
+      "step": 2838
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.312764127097154e-06,
+      "loss": 1.1311,
+      "step": 2839
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.30614652774547e-06,
+      "loss": 1.1827,
+      "step": 2840
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.29952969200004e-06,
+      "loss": 1.0691,
+      "step": 2841
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.292913622843803e-06,
+      "loss": 1.0397,
+      "step": 2842
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.286298323259348e-06,
+      "loss": 1.3287,
+      "step": 2843
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.279683796228919e-06,
+      "loss": 1.1102,
+      "step": 2844
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.273070044734415e-06,
+      "loss": 1.261,
+      "step": 2845
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.266457071757377e-06,
+      "loss": 1.0476,
+      "step": 2846
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.259844880279e-06,
+      "loss": 1.3238,
+      "step": 2847
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.253233473280135e-06,
+      "loss": 1.2017,
+      "step": 2848
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.24662285374126e-06,
+      "loss": 1.0488,
+      "step": 2849
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.24001302464251e-06,
+      "loss": 1.2034,
+      "step": 2850
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.233403988963675e-06,
+      "loss": 1.3877,
+      "step": 2851
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.226795749684162e-06,
+      "loss": 1.2408,
+      "step": 2852
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.220188309783036e-06,
+      "loss": 1.0695,
+      "step": 2853
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.213581672239001e-06,
+      "loss": 1.1931,
+      "step": 2854
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.206975840030392e-06,
+      "loss": 1.331,
+      "step": 2855
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.200370816135188e-06,
+      "loss": 1.3431,
+      "step": 2856
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.193766603530995e-06,
+      "loss": 1.2536,
+      "step": 2857
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.187163205195062e-06,
+      "loss": 1.1097,
+      "step": 2858
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.180560624104275e-06,
+      "loss": 0.9581,
+      "step": 2859
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.17395886323513e-06,
+      "loss": 1.4753,
+      "step": 2860
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.167357925563775e-06,
+      "loss": 1.1402,
+      "step": 2861
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.160757814065985e-06,
+      "loss": 1.2903,
+      "step": 2862
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.154158531717146e-06,
+      "loss": 0.9265,
+      "step": 2863
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.147560081492288e-06,
+      "loss": 1.3935,
+      "step": 2864
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.140962466366058e-06,
+      "loss": 1.0368,
+      "step": 2865
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.134365689312725e-06,
+      "loss": 1.0744,
+      "step": 2866
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.127769753306182e-06,
+      "loss": 1.4066,
+      "step": 2867
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.121174661319952e-06,
+      "loss": 1.0508,
+      "step": 2868
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.11458041632716e-06,
+      "loss": 1.4228,
+      "step": 2869
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.107987021300562e-06,
+      "loss": 1.0113,
+      "step": 2870
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.101394479212529e-06,
+      "loss": 1.0615,
+      "step": 2871
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.094802793035039e-06,
+      "loss": 1.3232,
+      "step": 2872
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.088211965739694e-06,
+      "loss": 1.167,
+      "step": 2873
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.081622000297713e-06,
+      "loss": 1.3217,
+      "step": 2874
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.075032899679906e-06,
+      "loss": 1.2542,
+      "step": 2875
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.068444666856714e-06,
+      "loss": 1.3509,
+      "step": 2876
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.06185730479818e-06,
+      "loss": 1.1499,
+      "step": 2877
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.055270816473949e-06,
+      "loss": 1.1683,
+      "step": 2878
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.048685204853275e-06,
+      "loss": 1.1711,
+      "step": 2879
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.042100472905026e-06,
+      "loss": 1.1933,
+      "step": 2880
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.035516623597658e-06,
+      "loss": 1.25,
+      "step": 2881
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.028933659899241e-06,
+      "loss": 1.5629,
+      "step": 2882
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.022351584777444e-06,
+      "loss": 1.163,
+      "step": 2883
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.015770401199524e-06,
+      "loss": 1.3876,
+      "step": 2884
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.009190112132355e-06,
+      "loss": 1.0828,
+      "step": 2885
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.00261072054239e-06,
+      "loss": 1.1576,
+      "step": 2886
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.996032229395686e-06,
+      "loss": 1.2027,
+      "step": 2887
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.9894546416579e-06,
+      "loss": 1.2683,
+      "step": 2888
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.98287796029427e-06,
+      "loss": 1.2397,
+      "step": 2889
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.976302188269628e-06,
+      "loss": 0.962,
+      "step": 2890
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.969727328548405e-06,
+      "loss": 1.2215,
+      "step": 2891
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.963153384094607e-06,
+      "loss": 1.2532,
+      "step": 2892
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.95658035787184e-06,
+      "loss": 1.1197,
+      "step": 2893
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.950008252843292e-06,
+      "loss": 1.2762,
+      "step": 2894
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.943437071971728e-06,
+      "loss": 1.242,
+      "step": 2895
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.936866818219504e-06,
+      "loss": 1.0811,
+      "step": 2896
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.930297494548563e-06,
+      "loss": 1.3726,
+      "step": 2897
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.923729103920415e-06,
+      "loss": 1.2735,
+      "step": 2898
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.91716164929616e-06,
+      "loss": 1.1131,
+      "step": 2899
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.910595133636474e-06,
+      "loss": 1.1898,
+      "step": 2900
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.904029559901604e-06,
+      "loss": 0.9982,
+      "step": 2901
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.897464931051384e-06,
+      "loss": 1.1719,
+      "step": 2902
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.890901250045212e-06,
+      "loss": 1.2704,
+      "step": 2903
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.884338519842059e-06,
+      "loss": 1.0922,
+      "step": 2904
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.877776743400472e-06,
+      "loss": 1.2431,
+      "step": 2905
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.87121592367857e-06,
+      "loss": 1.2905,
+      "step": 2906
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.864656063634029e-06,
+      "loss": 1.0458,
+      "step": 2907
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.858097166224104e-06,
+      "loss": 0.9877,
+      "step": 2908
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.851539234405614e-06,
+      "loss": 1.3599,
+      "step": 2909
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.844982271134939e-06,
+      "loss": 1.3973,
+      "step": 2910
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.838426279368026e-06,
+      "loss": 1.2347,
+      "step": 2911
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.831871262060377e-06,
+      "loss": 1.3039,
+      "step": 2912
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.825317222167067e-06,
+      "loss": 1.2116,
+      "step": 2913
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.818764162642722e-06,
+      "loss": 1.2052,
+      "step": 2914
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.812212086441522e-06,
+      "loss": 1.1111,
+      "step": 2915
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.805660996517213e-06,
+      "loss": 1.3294,
+      "step": 2916
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.799110895823097e-06,
+      "loss": 1.3044,
+      "step": 2917
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.792561787312016e-06,
+      "loss": 1.1495,
+      "step": 2918
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.78601367393638e-06,
+      "loss": 1.0134,
+      "step": 2919
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.779466558648145e-06,
+      "loss": 1.2259,
+      "step": 2920
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.772920444398815e-06,
+      "loss": 0.9717,
+      "step": 2921
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.766375334139445e-06,
+      "loss": 1.0057,
+      "step": 2922
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.759831230820637e-06,
+      "loss": 1.2673,
+      "step": 2923
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.753288137392538e-06,
+      "loss": 1.2386,
+      "step": 2924
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.74674605680484e-06,
+      "loss": 1.1085,
+      "step": 2925
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.740204992006787e-06,
+      "loss": 1.0794,
+      "step": 2926
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.733664945947144e-06,
+      "loss": 1.2609,
+      "step": 2927
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.727125921574236e-06,
+      "loss": 1.2857,
+      "step": 2928
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.720587921835927e-06,
+      "loss": 1.1709,
+      "step": 2929
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.714050949679603e-06,
+      "loss": 1.0896,
+      "step": 2930
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.707515008052201e-06,
+      "loss": 1.1397,
+      "step": 2931
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.700980099900196e-06,
+      "loss": 1.1293,
+      "step": 2932
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.69444622816958e-06,
+      "loss": 1.1206,
+      "step": 2933
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.687913395805893e-06,
+      "loss": 1.0803,
+      "step": 2934
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.681381605754205e-06,
+      "loss": 1.0567,
+      "step": 2935
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.674850860959109e-06,
+      "loss": 1.3034,
+      "step": 2936
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.66832116436473e-06,
+      "loss": 1.075,
+      "step": 2937
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.66179251891473e-06,
+      "loss": 1.2792,
+      "step": 2938
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.655264927552275e-06,
+      "loss": 0.995,
+      "step": 2939
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.648738393220078e-06,
+      "loss": 1.3361,
+      "step": 2940
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.642212918860361e-06,
+      "loss": 0.903,
+      "step": 2941
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.635688507414878e-06,
+      "loss": 1.002,
+      "step": 2942
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.629165161824901e-06,
+      "loss": 1.1119,
+      "step": 2943
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.622642885031214e-06,
+      "loss": 1.1531,
+      "step": 2944
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.616121679974129e-06,
+      "loss": 1.127,
+      "step": 2945
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.609601549593476e-06,
+      "loss": 1.1922,
+      "step": 2946
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.603082496828583e-06,
+      "loss": 1.1642,
+      "step": 2947
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.596564524618315e-06,
+      "loss": 1.2832,
+      "step": 2948
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.590047635901043e-06,
+      "loss": 1.2951,
+      "step": 2949
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.5835318336146334e-06,
+      "loss": 1.2993,
+      "step": 2950
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.577017120696484e-06,
+      "loss": 1.1907,
+      "step": 2951
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.570503500083494e-06,
+      "loss": 1.1713,
+      "step": 2952
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.563990974712065e-06,
+      "loss": 1.009,
+      "step": 2953
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.557479547518109e-06,
+      "loss": 1.3566,
+      "step": 2954
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.5509692214370515e-06,
+      "loss": 1.3784,
+      "step": 2955
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.544459999403802e-06,
+      "loss": 1.333,
+      "step": 2956
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.537951884352791e-06,
+      "loss": 1.1916,
+      "step": 2957
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.531444879217944e-06,
+      "loss": 1.2838,
+      "step": 2958
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.524938986932674e-06,
+      "loss": 1.1554,
+      "step": 2959
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.5184342104299125e-06,
+      "loss": 1.0584,
+      "step": 2960
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.5119305526420805e-06,
+      "loss": 1.0896,
+      "step": 2961
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.505428016501083e-06,
+      "loss": 1.2635,
+      "step": 2962
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.4989266049383345e-06,
+      "loss": 1.1452,
+      "step": 2963
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.492426320884738e-06,
+      "loss": 1.1994,
+      "step": 2964
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.485927167270684e-06,
+      "loss": 1.38,
+      "step": 2965
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.47942914702606e-06,
+      "loss": 1.1249,
+      "step": 2966
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.472932263080237e-06,
+      "loss": 0.9212,
+      "step": 2967
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.466436518362076e-06,
+      "loss": 1.2933,
+      "step": 2968
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.459941915799929e-06,
+      "loss": 1.2321,
+      "step": 2969
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.453448458321616e-06,
+      "loss": 1.0211,
+      "step": 2970
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.4469561488544654e-06,
+      "loss": 1.2227,
+      "step": 2971
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.440464990325275e-06,
+      "loss": 1.2709,
+      "step": 2972
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.433974985660317e-06,
+      "loss": 1.0309,
+      "step": 2973
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.427486137785357e-06,
+      "loss": 1.2957,
+      "step": 2974
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.4209984496256325e-06,
+      "loss": 1.0213,
+      "step": 2975
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.414511924105856e-06,
+      "loss": 1.2223,
+      "step": 2976
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.408026564150222e-06,
+      "loss": 1.0976,
+      "step": 2977
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.401542372682398e-06,
+      "loss": 1.0764,
+      "step": 2978
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.395059352625516e-06,
+      "loss": 1.2684,
+      "step": 2979
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.38857750690219e-06,
+      "loss": 1.1654,
+      "step": 2980
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.382096838434509e-06,
+      "loss": 1.4128,
+      "step": 2981
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.375617350144011e-06,
+      "loss": 1.0498,
+      "step": 2982
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.369139044951723e-06,
+      "loss": 1.2307,
+      "step": 2983
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.3626619257781316e-06,
+      "loss": 1.2365,
+      "step": 2984
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.356185995543181e-06,
+      "loss": 1.1533,
+      "step": 2985
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.349711257166291e-06,
+      "loss": 1.2592,
+      "step": 2986
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.343237713566338e-06,
+      "loss": 1.0819,
+      "step": 2987
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.336765367661659e-06,
+      "loss": 1.2872,
+      "step": 2988
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.330294222370054e-06,
+      "loss": 1.0124,
+      "step": 2989
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.323824280608782e-06,
+      "loss": 1.1859,
+      "step": 2990
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.317355545294558e-06,
+      "loss": 1.222,
+      "step": 2991
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.310888019343549e-06,
+      "loss": 1.1435,
+      "step": 2992
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.304421705671391e-06,
+      "loss": 1.2372,
+      "step": 2993
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.297956607193148e-06,
+      "loss": 0.9663,
+      "step": 2994
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.291492726823365e-06,
+      "loss": 1.0927,
+      "step": 2995
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.285030067476024e-06,
+      "loss": 1.16,
+      "step": 2996
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.278568632064548e-06,
+      "loss": 1.1493,
+      "step": 2997
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.272108423501825e-06,
+      "loss": 1.1034,
+      "step": 2998
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.265649444700178e-06,
+      "loss": 1.3845,
+      "step": 2999
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.259191698571381e-06,
+      "loss": 1.0733,
+      "step": 3000
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.252735188026655e-06,
+      "loss": 1.2328,
+      "step": 3001
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.246279915976655e-06,
+      "loss": 1.191,
+      "step": 3002
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.239825885331484e-06,
+      "loss": 1.0239,
+      "step": 3003
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.23337309900069e-06,
+      "loss": 1.3123,
+      "step": 3004
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.2269215598932455e-06,
+      "loss": 1.2504,
+      "step": 3005
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.22047127091757e-06,
+      "loss": 1.2093,
+      "step": 3006
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.214022234981531e-06,
+      "loss": 1.2885,
+      "step": 3007
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.207574454992403e-06,
+      "loss": 1.2141,
+      "step": 3008
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.20112793385692e-06,
+      "loss": 1.3203,
+      "step": 3009
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.19468267448124e-06,
+      "loss": 1.0908,
+      "step": 3010
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.1882386797709426e-06,
+      "loss": 0.964,
+      "step": 3011
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.181795952631052e-06,
+      "loss": 1.2772,
+      "step": 3012
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.1753544959660145e-06,
+      "loss": 1.1213,
+      "step": 3013
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.168914312679702e-06,
+      "loss": 1.1582,
+      "step": 3014
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.162475405675414e-06,
+      "loss": 1.2427,
+      "step": 3015
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.15603777785588e-06,
+      "loss": 1.1981,
+      "step": 3016
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.14960143212324e-06,
+      "loss": 1.118,
+      "step": 3017
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.143166371379065e-06,
+      "loss": 1.1957,
+      "step": 3018
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.136732598524352e-06,
+      "loss": 1.1852,
+      "step": 3019
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.130300116459503e-06,
+      "loss": 0.9342,
+      "step": 3020
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.12386892808435e-06,
+      "loss": 1.1614,
+      "step": 3021
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.117439036298137e-06,
+      "loss": 1.3341,
+      "step": 3022
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.111010443999524e-06,
+      "loss": 0.9674,
+      "step": 3023
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.104583154086588e-06,
+      "loss": 1.1276,
+      "step": 3024
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.0981571694568074e-06,
+      "loss": 1.0549,
+      "step": 3025
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.091732493007091e-06,
+      "loss": 1.2335,
+      "step": 3026
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.085309127633746e-06,
+      "loss": 1.0433,
+      "step": 3027
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.0788870762324845e-06,
+      "loss": 1.1888,
+      "step": 3028
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.0724663416984365e-06,
+      "loss": 1.0099,
+      "step": 3029
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.066046926926136e-06,
+      "loss": 1.1855,
+      "step": 3030
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.059628834809514e-06,
+      "loss": 1.3001,
+      "step": 3031
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.053212068241916e-06,
+      "loss": 1.2313,
+      "step": 3032
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.046796630116085e-06,
+      "loss": 1.2712,
+      "step": 3033
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.040382523324163e-06,
+      "loss": 1.4764,
+      "step": 3034
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.033969750757694e-06,
+      "loss": 1.2846,
+      "step": 3035
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.027558315307628e-06,
+      "loss": 1.3558,
+      "step": 3036
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.021148219864292e-06,
+      "loss": 1.2017,
+      "step": 3037
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.014739467317431e-06,
+      "loss": 1.1406,
+      "step": 3038
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.008332060556176e-06,
+      "loss": 1.3797,
+      "step": 3039
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.001926002469045e-06,
+      "loss": 1.1214,
+      "step": 3040
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.9955212959439564e-06,
+      "loss": 1.3084,
+      "step": 3041
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.989117943868217e-06,
+      "loss": 0.9861,
+      "step": 3042
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.982715949128522e-06,
+      "loss": 1.3979,
+      "step": 3043
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.9763153146109554e-06,
+      "loss": 1.2469,
+      "step": 3044
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.96991604320099e-06,
+      "loss": 1.0383,
+      "step": 3045
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.963518137783477e-06,
+      "loss": 1.5185,
+      "step": 3046
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.95712160124266e-06,
+      "loss": 1.3029,
+      "step": 3047
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.950726436462167e-06,
+      "loss": 1.4701,
+      "step": 3048
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.944332646324992e-06,
+      "loss": 1.0003,
+      "step": 3049
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.937940233713527e-06,
+      "loss": 1.2394,
+      "step": 3050
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.93154920150954e-06,
+      "loss": 0.9721,
+      "step": 3051
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.925159552594165e-06,
+      "loss": 1.2941,
+      "step": 3052
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.918771289847924e-06,
+      "loss": 1.313,
+      "step": 3053
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.912384416150708e-06,
+      "loss": 1.0863,
+      "step": 3054
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.905998934381788e-06,
+      "loss": 1.1421,
+      "step": 3055
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.899614847419802e-06,
+      "loss": 1.0898,
+      "step": 3056
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.893232158142758e-06,
+      "loss": 1.2898,
+      "step": 3057
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.886850869428041e-06,
+      "loss": 1.0617,
+      "step": 3058
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.880470984152402e-06,
+      "loss": 1.1987,
+      "step": 3059
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.874092505191951e-06,
+      "loss": 1.127,
+      "step": 3060
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.867715435422171e-06,
+      "loss": 1.5348,
+      "step": 3061
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.86133977771792e-06,
+      "loss": 1.2778,
+      "step": 3062
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.8549655349533976e-06,
+      "loss": 0.8998,
+      "step": 3063
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.848592710002181e-06,
+      "loss": 1.2248,
+      "step": 3064
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.842221305737206e-06,
+      "loss": 1.1647,
+      "step": 3065
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.835851325030761e-06,
+      "loss": 1.0132,
+      "step": 3066
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.829482770754501e-06,
+      "loss": 1.246,
+      "step": 3067
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.823115645779437e-06,
+      "loss": 1.1652,
+      "step": 3068
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.816749952975927e-06,
+      "loss": 1.3744,
+      "step": 3069
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.810385695213695e-06,
+      "loss": 1.2964,
+      "step": 3070
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.804022875361813e-06,
+      "loss": 1.0873,
+      "step": 3071
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.797661496288699e-06,
+      "loss": 1.3435,
+      "step": 3072
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.791301560862128e-06,
+      "loss": 0.9776,
+      "step": 3073
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.784943071949232e-06,
+      "loss": 1.6198,
+      "step": 3074
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.77858603241647e-06,
+      "loss": 1.2663,
+      "step": 3075
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.772230445129666e-06,
+      "loss": 1.3549,
+      "step": 3076
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.765876312953983e-06,
+      "loss": 1.1743,
+      "step": 3077
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.759523638753925e-06,
+      "loss": 1.0105,
+      "step": 3078
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.753172425393344e-06,
+      "loss": 1.1266,
+      "step": 3079
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.746822675735433e-06,
+      "loss": 1.1814,
+      "step": 3080
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.740474392642719e-06,
+      "loss": 1.172,
+      "step": 3081
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.734127578977077e-06,
+      "loss": 1.2633,
+      "step": 3082
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.7277822375997095e-06,
+      "loss": 1.2785,
+      "step": 3083
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.7214383713711615e-06,
+      "loss": 1.2099,
+      "step": 3084
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.715095983151316e-06,
+      "loss": 1.5061,
+      "step": 3085
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.70875507579938e-06,
+      "loss": 1.1901,
+      "step": 3086
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.7024156521739035e-06,
+      "loss": 1.1807,
+      "step": 3087
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.6960777151327604e-06,
+      "loss": 1.0845,
+      "step": 3088
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.689741267533158e-06,
+      "loss": 1.2826,
+      "step": 3089
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.683406312231628e-06,
+      "loss": 1.2637,
+      "step": 3090
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.677072852084037e-06,
+      "loss": 1.2797,
+      "step": 3091
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.670740889945567e-06,
+      "loss": 1.2113,
+      "step": 3092
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.664410428670734e-06,
+      "loss": 1.0583,
+      "step": 3093
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.658081471113378e-06,
+      "loss": 1.0898,
+      "step": 3094
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.6517540201266484e-06,
+      "loss": 1.0209,
+      "step": 3095
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.645428078563027e-06,
+      "loss": 1.3103,
+      "step": 3096
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.639103649274316e-06,
+      "loss": 1.3259,
+      "step": 3097
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.632780735111628e-06,
+      "loss": 1.2806,
+      "step": 3098
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.6264593389253976e-06,
+      "loss": 1.1127,
+      "step": 3099
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.620139463565376e-06,
+      "loss": 1.0762,
+      "step": 3100
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.613821111880623e-06,
+      "loss": 1.2824,
+      "step": 3101
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.6075042867195176e-06,
+      "loss": 1.4929,
+      "step": 3102
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.60118899092975e-06,
+      "loss": 1.1881,
+      "step": 3103
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.594875227358315e-06,
+      "loss": 1.2698,
+      "step": 3104
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.5885629988515245e-06,
+      "loss": 1.1335,
+      "step": 3105
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.5822523082549974e-06,
+      "loss": 0.9732,
+      "step": 3106
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.5759431584136495e-06,
+      "loss": 1.2672,
+      "step": 3107
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.569635552171715e-06,
+      "loss": 1.0595,
+      "step": 3108
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.5633294923727235e-06,
+      "loss": 1.2652,
+      "step": 3109
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.55702498185951e-06,
+      "loss": 1.0307,
+      "step": 3110
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.550722023474216e-06,
+      "loss": 1.1597,
+      "step": 3111
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.5444206200582736e-06,
+      "loss": 1.224,
+      "step": 3112
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.538120774452423e-06,
+      "loss": 1.2247,
+      "step": 3113
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.531822489496699e-06,
+      "loss": 1.3165,
+      "step": 3114
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.525525768030423e-06,
+      "loss": 1.4738,
+      "step": 3115
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.519230612892231e-06,
+      "loss": 1.1606,
+      "step": 3116
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.512937026920043e-06,
+      "loss": 1.0181,
+      "step": 3117
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.506645012951062e-06,
+      "loss": 1.1831,
+      "step": 3118
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.500354573821796e-06,
+      "loss": 1.2707,
+      "step": 3119
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.494065712368043e-06,
+      "loss": 1.3089,
+      "step": 3120
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.487778431424878e-06,
+      "loss": 1.1474,
+      "step": 3121
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.481492733826673e-06,
+      "loss": 0.9664,
+      "step": 3122
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.475208622407089e-06,
+      "loss": 1.1036,
+      "step": 3123
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.46892609999906e-06,
+      "loss": 1.2544,
+      "step": 3124
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.462645169434814e-06,
+      "loss": 1.2601,
+      "step": 3125
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.45636583354586e-06,
+      "loss": 1.1798,
+      "step": 3126
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.4500880951629785e-06,
+      "loss": 1.2801,
+      "step": 3127
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.4438119571162435e-06,
+      "loss": 1.2247,
+      "step": 3128
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.437537422235005e-06,
+      "loss": 1.0776,
+      "step": 3129
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.431264493347878e-06,
+      "loss": 1.2495,
+      "step": 3130
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.424993173282765e-06,
+      "loss": 1.313,
+      "step": 3131
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.418723464866845e-06,
+      "loss": 0.8711,
+      "step": 3132
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.4124553709265606e-06,
+      "loss": 1.1932,
+      "step": 3133
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.406188894287632e-06,
+      "loss": 1.1519,
+      "step": 3134
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.399924037775056e-06,
+      "loss": 1.2469,
+      "step": 3135
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.393660804213086e-06,
+      "loss": 1.0836,
+      "step": 3136
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.387399196425259e-06,
+      "loss": 1.1073,
+      "step": 3137
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.381139217234361e-06,
+      "loss": 1.1013,
+      "step": 3138
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.374880869462458e-06,
+      "loss": 1.0856,
+      "step": 3139
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.368624155930882e-06,
+      "loss": 1.0808,
+      "step": 3140
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.3623690794602155e-06,
+      "loss": 1.3334,
+      "step": 3141
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.356115642870311e-06,
+      "loss": 1.0975,
+      "step": 3142
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.349863848980286e-06,
+      "loss": 1.3674,
+      "step": 3143
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.343613700608507e-06,
+      "loss": 1.0312,
+      "step": 3144
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.337365200572604e-06,
+      "loss": 1.2751,
+      "step": 3145
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.331118351689469e-06,
+      "loss": 1.1696,
+      "step": 3146
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.32487315677524e-06,
+      "loss": 0.9051,
+      "step": 3147
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.318629618645314e-06,
+      "loss": 1.0745,
+      "step": 3148
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.312387740114348e-06,
+      "loss": 1.0542,
+      "step": 3149
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.306147523996233e-06,
+      "loss": 1.0619,
+      "step": 3150
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.299908973104125e-06,
+      "loss": 1.099,
+      "step": 3151
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.293672090250434e-06,
+      "loss": 1.071,
+      "step": 3152
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.2874368782467994e-06,
+      "loss": 1.0885,
+      "step": 3153
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.281203339904122e-06,
+      "loss": 1.2799,
+      "step": 3154
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.274971478032546e-06,
+      "loss": 1.4421,
+      "step": 3155
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.268741295441452e-06,
+      "loss": 1.0088,
+      "step": 3156
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.262512794939472e-06,
+      "loss": 1.2061,
+      "step": 3157
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.2562859793344795e-06,
+      "loss": 1.3309,
+      "step": 3158
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.2500608514335835e-06,
+      "loss": 1.2397,
+      "step": 3159
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.243837414043133e-06,
+      "loss": 1.2476,
+      "step": 3160
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.237615669968724e-06,
+      "loss": 1.1051,
+      "step": 3161
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.231395622015171e-06,
+      "loss": 1.2678,
+      "step": 3162
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.225177272986536e-06,
+      "loss": 1.207,
+      "step": 3163
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.2189606256861245e-06,
+      "loss": 1.0772,
+      "step": 3164
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.2127456829164525e-06,
+      "loss": 0.978,
+      "step": 3165
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.206532447479283e-06,
+      "loss": 1.1461,
+      "step": 3166
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.200320922175604e-06,
+      "loss": 1.1632,
+      "step": 3167
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.194111109805635e-06,
+      "loss": 1.2474,
+      "step": 3168
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.187903013168826e-06,
+      "loss": 1.187,
+      "step": 3169
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.181696635063839e-06,
+      "loss": 1.2241,
+      "step": 3170
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.1754919782885815e-06,
+      "loss": 1.2564,
+      "step": 3171
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.169289045640177e-06,
+      "loss": 1.199,
+      "step": 3172
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.163087839914961e-06,
+      "loss": 1.1222,
+      "step": 3173
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.156888363908504e-06,
+      "loss": 1.1262,
+      "step": 3174
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.1506906204155956e-06,
+      "loss": 1.0835,
+      "step": 3175
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.144494612230236e-06,
+      "loss": 1.1547,
+      "step": 3176
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.1383003421456515e-06,
+      "loss": 1.2947,
+      "step": 3177
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.132107812954281e-06,
+      "loss": 1.2394,
+      "step": 3178
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.125917027447779e-06,
+      "loss": 1.1705,
+      "step": 3179
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.1197279884170125e-06,
+      "loss": 1.0179,
+      "step": 3180
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.11354069865207e-06,
+      "loss": 1.3331,
+      "step": 3181
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.107355160942232e-06,
+      "loss": 1.1835,
+      "step": 3182
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.10117137807601e-06,
+      "loss": 1.357,
+      "step": 3183
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.0949893528411185e-06,
+      "loss": 1.2572,
+      "step": 3184
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.088809088024467e-06,
+      "loss": 1.0603,
+      "step": 3185
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.082630586412188e-06,
+      "loss": 1.0638,
+      "step": 3186
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.076453850789612e-06,
+      "loss": 1.1982,
+      "step": 3187
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.070278883941272e-06,
+      "loss": 0.9647,
+      "step": 3188
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.064105688650905e-06,
+      "loss": 1.2992,
+      "step": 3189
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.057934267701455e-06,
+      "loss": 1.0607,
+      "step": 3190
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.051764623875054e-06,
+      "loss": 1.3923,
+      "step": 3191
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.045596759953042e-06,
+      "loss": 1.2177,
+      "step": 3192
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.0394306787159595e-06,
+      "loss": 1.0235,
+      "step": 3193
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.0332663829435275e-06,
+      "loss": 1.234,
+      "step": 3194
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.027103875414686e-06,
+      "loss": 1.4622,
+      "step": 3195
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.020943158907544e-06,
+      "loss": 1.2598,
+      "step": 3196
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.01478423619942e-06,
+      "loss": 1.3247,
+      "step": 3197
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.0086271100668206e-06,
+      "loss": 1.1769,
+      "step": 3198
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.002471783285435e-06,
+      "loss": 1.2148,
+      "step": 3199
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.99631825863015e-06,
+      "loss": 1.2193,
+      "step": 3200
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.990166538875038e-06,
+      "loss": 1.1524,
+      "step": 3201
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.984016626793352e-06,
+      "loss": 1.295,
+      "step": 3202
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.977868525157539e-06,
+      "loss": 1.0274,
+      "step": 3203
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.971722236739226e-06,
+      "loss": 1.1832,
+      "step": 3204
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.9655777643092165e-06,
+      "loss": 1.1507,
+      "step": 3205
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.9594351106375005e-06,
+      "loss": 1.0654,
+      "step": 3206
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.953294278493258e-06,
+      "loss": 1.1603,
+      "step": 3207
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.947155270644828e-06,
+      "loss": 1.1986,
+      "step": 3208
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.94101808985974e-06,
+      "loss": 1.2483,
+      "step": 3209
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.9348827389047e-06,
+      "loss": 1.323,
+      "step": 3210
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.928749220545582e-06,
+      "loss": 1.4486,
+      "step": 3211
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.922617537547442e-06,
+      "loss": 1.2488,
+      "step": 3212
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.916487692674503e-06,
+      "loss": 1.3225,
+      "step": 3213
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.9103596886901596e-06,
+      "loss": 1.0904,
+      "step": 3214
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.9042335283569785e-06,
+      "loss": 1.2936,
+      "step": 3215
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.898109214436701e-06,
+      "loss": 1.1819,
+      "step": 3216
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.891986749690217e-06,
+      "loss": 1.0891,
+      "step": 3217
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.885866136877603e-06,
+      "loss": 1.1053,
+      "step": 3218
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.879747378758096e-06,
+      "loss": 1.1876,
+      "step": 3219
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.8736304780900885e-06,
+      "loss": 0.9801,
+      "step": 3220
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.867515437631144e-06,
+      "loss": 1.2452,
+      "step": 3221
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.861402260137983e-06,
+      "loss": 1.4216,
+      "step": 3222
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.855290948366487e-06,
+      "loss": 1.197,
+      "step": 3223
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.8491815050717e-06,
+      "loss": 1.2724,
+      "step": 3224
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.843073933007817e-06,
+      "loss": 0.9947,
+      "step": 3225
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.8369682349281964e-06,
+      "loss": 1.0089,
+      "step": 3226
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.830864413585349e-06,
+      "loss": 1.2467,
+      "step": 3227
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.824762471730937e-06,
+      "loss": 1.1102,
+      "step": 3228
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.8186624121157745e-06,
+      "loss": 1.1121,
+      "step": 3229
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.812564237489837e-06,
+      "loss": 1.0364,
+      "step": 3230
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.806467950602239e-06,
+      "loss": 1.0756,
+      "step": 3231
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.8003735542012465e-06,
+      "loss": 1.2072,
+      "step": 3232
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.794281051034285e-06,
+      "loss": 1.3854,
+      "step": 3233
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.7881904438478995e-06,
+      "loss": 1.0788,
+      "step": 3234
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.782101735387811e-06,
+      "loss": 1.2107,
+      "step": 3235
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.776014928398867e-06,
+      "loss": 1.0754,
+      "step": 3236
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.7699300256250565e-06,
+      "loss": 1.0317,
+      "step": 3237
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.763847029809524e-06,
+      "loss": 1.1865,
+      "step": 3238
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.75776594369454e-06,
+      "loss": 1.156,
+      "step": 3239
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.751686770021518e-06,
+      "loss": 1.0716,
+      "step": 3240
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.745609511531016e-06,
+      "loss": 1.19,
+      "step": 3241
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.7395341709627215e-06,
+      "loss": 1.3901,
+      "step": 3242
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.733460751055455e-06,
+      "loss": 1.1099,
+      "step": 3243
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.727389254547182e-06,
+      "loss": 1.2431,
+      "step": 3244
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.721319684174994e-06,
+      "loss": 1.1954,
+      "step": 3245
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.715252042675111e-06,
+      "loss": 1.1277,
+      "step": 3246
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.709186332782886e-06,
+      "loss": 1.1925,
+      "step": 3247
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.703122557232808e-06,
+      "loss": 1.3295,
+      "step": 3248
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.697060718758485e-06,
+      "loss": 1.12,
+      "step": 3249
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.691000820092655e-06,
+      "loss": 1.3481,
+      "step": 3250
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.6849428639671775e-06,
+      "loss": 1.0231,
+      "step": 3251
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.678886853113048e-06,
+      "loss": 1.0317,
+      "step": 3252
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.672832790260375e-06,
+      "loss": 1.0383,
+      "step": 3253
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.666780678138386e-06,
+      "loss": 1.4779,
+      "step": 3254
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.660730519475442e-06,
+      "loss": 1.0639,
+      "step": 3255
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.654682316999012e-06,
+      "loss": 1.2453,
+      "step": 3256
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.648636073435687e-06,
+      "loss": 1.1676,
+      "step": 3257
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.6425917915111695e-06,
+      "loss": 1.4223,
+      "step": 3258
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.6365494739502926e-06,
+      "loss": 1.4467,
+      "step": 3259
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.6305091234769895e-06,
+      "loss": 1.1884,
+      "step": 3260
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.624470742814306e-06,
+      "loss": 1.1055,
+      "step": 3261
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.618434334684415e-06,
+      "loss": 0.9597,
+      "step": 3262
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.6123999018085855e-06,
+      "loss": 1.229,
+      "step": 3263
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.606367446907195e-06,
+      "loss": 1.191,
+      "step": 3264
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.6003369726997494e-06,
+      "loss": 1.224,
+      "step": 3265
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.594308481904831e-06,
+      "loss": 1.2176,
+      "step": 3266
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.5882819772401504e-06,
+      "loss": 1.1913,
+      "step": 3267
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.5822574614225285e-06,
+      "loss": 1.1641,
+      "step": 3268
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.576234937167859e-06,
+      "loss": 1.3452,
+      "step": 3269
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.57021440719117e-06,
+      "loss": 1.0332,
+      "step": 3270
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.564195874206573e-06,
+      "loss": 1.3675,
+      "step": 3271
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.558179340927279e-06,
+      "loss": 1.0586,
+      "step": 3272
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.552164810065609e-06,
+      "loss": 1.193,
+      "step": 3273
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.546152284332972e-06,
+      "loss": 1.0776,
+      "step": 3274
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.540141766439871e-06,
+      "loss": 1.1495,
+      "step": 3275
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.534133259095915e-06,
+      "loss": 1.17,
+      "step": 3276
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.528126765009798e-06,
+      "loss": 1.2231,
+      "step": 3277
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.522122286889302e-06,
+      "loss": 1.1482,
+      "step": 3278
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.51611982744132e-06,
+      "loss": 1.3223,
+      "step": 3279
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.510119389371804e-06,
+      "loss": 1.1958,
+      "step": 3280
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.504120975385825e-06,
+      "loss": 1.2034,
+      "step": 3281
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.498124588187525e-06,
+      "loss": 1.0416,
+      "step": 3282
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.4921302304801305e-06,
+      "loss": 1.3152,
+      "step": 3283
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.486137904965968e-06,
+      "loss": 1.0164,
+      "step": 3284
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.4801476143464335e-06,
+      "loss": 1.3494,
+      "step": 3285
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.474159361322007e-06,
+      "loss": 1.0174,
+      "step": 3286
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.46817314859226e-06,
+      "loss": 1.1294,
+      "step": 3287
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.462188978855836e-06,
+      "loss": 1.0793,
+      "step": 3288
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.456206854810457e-06,
+      "loss": 0.9713,
+      "step": 3289
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.450226779152923e-06,
+      "loss": 0.9312,
+      "step": 3290
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.444248754579118e-06,
+      "loss": 1.1429,
+      "step": 3291
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.438272783783994e-06,
+      "loss": 0.8471,
+      "step": 3292
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.432298869461574e-06,
+      "loss": 1.2415,
+      "step": 3293
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.426327014304968e-06,
+      "loss": 1.1586,
+      "step": 3294
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.420357221006344e-06,
+      "loss": 0.979,
+      "step": 3295
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.4143894922569404e-06,
+      "loss": 0.8567,
+      "step": 3296
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.408423830747078e-06,
+      "loss": 1.0661,
+      "step": 3297
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.402460239166134e-06,
+      "loss": 1.1202,
+      "step": 3298
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.396498720202551e-06,
+      "loss": 0.9591,
+      "step": 3299
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.390539276543854e-06,
+      "loss": 0.9783,
+      "step": 3300
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.384581910876605e-06,
+      "loss": 1.026,
+      "step": 3301
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.378626625886451e-06,
+      "loss": 1.0141,
+      "step": 3302
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.372673424258106e-06,
+      "loss": 0.984,
+      "step": 3303
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.366722308675313e-06,
+      "loss": 0.8774,
+      "step": 3304
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.36077328182091e-06,
+      "loss": 1.3057,
+      "step": 3305
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.354826346376773e-06,
+      "loss": 1.1682,
+      "step": 3306
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.348881505023839e-06,
+      "loss": 1.093,
+      "step": 3307
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.3429387604421115e-06,
+      "loss": 1.0644,
+      "step": 3308
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.3369981153106256e-06,
+      "loss": 1.155,
+      "step": 3309
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.33105957230749e-06,
+      "loss": 0.9048,
+      "step": 3310
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.325123134109871e-06,
+      "loss": 1.1777,
+      "step": 3311
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.319188803393957e-06,
+      "loss": 1.0806,
+      "step": 3312
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.313256582835016e-06,
+      "loss": 0.8797,
+      "step": 3313
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.307326475107349e-06,
+      "loss": 1.0386,
+      "step": 3314
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.301398482884307e-06,
+      "loss": 0.9501,
+      "step": 3315
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.295472608838293e-06,
+      "loss": 1.1863,
+      "step": 3316
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.289548855640749e-06,
+      "loss": 1.0759,
+      "step": 3317
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.2836272259621605e-06,
+      "loss": 1.0167,
+      "step": 3318
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.277707722472061e-06,
+      "loss": 1.0939,
+      "step": 3319
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.271790347839023e-06,
+      "loss": 0.9224,
+      "step": 3320
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.2658751047306525e-06,
+      "loss": 1.049,
+      "step": 3321
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.25996199581361e-06,
+      "loss": 0.9364,
+      "step": 3322
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.254051023753579e-06,
+      "loss": 0.8037,
+      "step": 3323
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.248142191215286e-06,
+      "loss": 1.0876,
+      "step": 3324
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.242235500862491e-06,
+      "loss": 0.8605,
+      "step": 3325
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.236330955357996e-06,
+      "loss": 1.0815,
+      "step": 3326
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.230428557363624e-06,
+      "loss": 0.9909,
+      "step": 3327
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.224528309540233e-06,
+      "loss": 0.973,
+      "step": 3328
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.218630214547723e-06,
+      "loss": 0.9271,
+      "step": 3329
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.2127342750450105e-06,
+      "loss": 0.9853,
+      "step": 3330
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.20684049369004e-06,
+      "loss": 0.9752,
+      "step": 3331
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.2009488731398e-06,
+      "loss": 0.8973,
+      "step": 3332
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.195059416050275e-06,
+      "loss": 0.9018,
+      "step": 3333
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.189172125076507e-06,
+      "loss": 1.221,
+      "step": 3334
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.1832870028725406e-06,
+      "loss": 1.0561,
+      "step": 3335
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.177404052091444e-06,
+      "loss": 0.9395,
+      "step": 3336
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.171523275385317e-06,
+      "loss": 1.0935,
+      "step": 3337
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.165644675405274e-06,
+      "loss": 1.006,
+      "step": 3338
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.159768254801439e-06,
+      "loss": 1.0451,
+      "step": 3339
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.1538940162229715e-06,
+      "loss": 1.0752,
+      "step": 3340
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.148021962318034e-06,
+      "loss": 1.2291,
+      "step": 3341
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.1421520957338025e-06,
+      "loss": 1.0072,
+      "step": 3342
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.136284419116484e-06,
+      "loss": 1.0638,
+      "step": 3343
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.130418935111272e-06,
+      "loss": 1.1057,
+      "step": 3344
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.124555646362392e-06,
+      "loss": 1.0071,
+      "step": 3345
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.118694555513084e-06,
+      "loss": 1.1144,
+      "step": 3346
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.112835665205569e-06,
+      "loss": 1.1128,
+      "step": 3347
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.106978978081108e-06,
+      "loss": 0.8351,
+      "step": 3348
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.101124496779948e-06,
+      "loss": 0.9286,
+      "step": 3349
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.095272223941346e-06,
+      "loss": 1.0874,
+      "step": 3350
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.089422162203572e-06,
+      "loss": 1.0874,
+      "step": 3351
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.0835743142038915e-06,
+      "loss": 1.018,
+      "step": 3352
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.07772868257857e-06,
+      "loss": 0.8934,
+      "step": 3353
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.071885269962882e-06,
+      "loss": 1.2366,
+      "step": 3354
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.066044078991095e-06,
+      "loss": 1.2655,
+      "step": 3355
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.060205112296476e-06,
+      "loss": 0.9573,
+      "step": 3356
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.054368372511287e-06,
+      "loss": 0.9512,
+      "step": 3357
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.048533862266795e-06,
+      "loss": 1.1683,
+      "step": 3358
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 5.042701584193257e-06,
+      "loss": 1.0006,
+      "step": 3359
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 5.036871540919914e-06,
+      "loss": 0.9658,
+      "step": 3360
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 5.031043735075019e-06,
+      "loss": 1.0607,
+      "step": 3361
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 5.025218169285799e-06,
+      "loss": 1.1248,
+      "step": 3362
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 5.0193948461784824e-06,
+      "loss": 0.9537,
+      "step": 3363
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 5.013573768378275e-06,
+      "loss": 0.8562,
+      "step": 3364
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 5.007754938509386e-06,
+      "loss": 1.148,
+      "step": 3365
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 5.001938359194997e-06,
+      "loss": 0.9852,
+      "step": 3366
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.9961240330572846e-06,
+      "loss": 1.1392,
+      "step": 3367
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.990311962717399e-06,
+      "loss": 1.1638,
+      "step": 3368
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.984502150795489e-06,
+      "loss": 1.0825,
+      "step": 3369
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.978694599910672e-06,
+      "loss": 0.9784,
+      "step": 3370
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.972889312681047e-06,
+      "loss": 1.1608,
+      "step": 3371
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.967086291723704e-06,
+      "loss": 1.1235,
+      "step": 3372
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.9612855396547e-06,
+      "loss": 0.9897,
+      "step": 3373
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.955487059089067e-06,
+      "loss": 0.8838,
+      "step": 3374
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.949690852640828e-06,
+      "loss": 1.1064,
+      "step": 3375
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.9438969229229685e-06,
+      "loss": 0.9198,
+      "step": 3376
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.938105272547444e-06,
+      "loss": 1.1383,
+      "step": 3377
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.932315904125202e-06,
+      "loss": 0.898,
+      "step": 3378
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.926528820266134e-06,
+      "loss": 1.0415,
+      "step": 3379
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.920744023579125e-06,
+      "loss": 1.0552,
+      "step": 3380
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.9149615166720166e-06,
+      "loss": 0.7998,
+      "step": 3381
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.9091813021516186e-06,
+      "loss": 0.8152,
+      "step": 3382
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.903403382623718e-06,
+      "loss": 1.0745,
+      "step": 3383
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.897627760693054e-06,
+      "loss": 0.9762,
+      "step": 3384
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.891854438963331e-06,
+      "loss": 1.0619,
+      "step": 3385
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.886083420037231e-06,
+      "loss": 1.189,
+      "step": 3386
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.880314706516383e-06,
+      "loss": 1.1098,
+      "step": 3387
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.874548301001375e-06,
+      "loss": 1.1034,
+      "step": 3388
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.86878420609177e-06,
+      "loss": 1.0854,
+      "step": 3389
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.863022424386078e-06,
+      "loss": 1.2451,
+      "step": 3390
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.857262958481767e-06,
+      "loss": 1.047,
+      "step": 3391
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.8515058109752625e-06,
+      "loss": 0.9833,
+      "step": 3392
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.84575098446194e-06,
+      "loss": 1.0014,
+      "step": 3393
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.839998481536142e-06,
+      "loss": 0.997,
+      "step": 3394
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.834248304791153e-06,
+      "loss": 1.0032,
+      "step": 3395
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.828500456819203e-06,
+      "loss": 1.1344,
+      "step": 3396
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.82275494021149e-06,
+      "loss": 1.0615,
+      "step": 3397
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.817011757558146e-06,
+      "loss": 0.878,
+      "step": 3398
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.811270911448255e-06,
+      "loss": 0.9745,
+      "step": 3399
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.805532404469851e-06,
+      "loss": 1.0996,
+      "step": 3400
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.7997962392099115e-06,
+      "loss": 0.9887,
+      "step": 3401
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.794062418254356e-06,
+      "loss": 1.0371,
+      "step": 3402
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.788330944188045e-06,
+      "loss": 0.976,
+      "step": 3403
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.782601819594792e-06,
+      "loss": 1.0322,
+      "step": 3404
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.776875047057341e-06,
+      "loss": 1.1872,
+      "step": 3405
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.771150629157376e-06,
+      "loss": 1.2163,
+      "step": 3406
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.765428568475529e-06,
+      "loss": 0.8217,
+      "step": 3407
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.759708867591359e-06,
+      "loss": 0.8608,
+      "step": 3408
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.753991529083361e-06,
+      "loss": 0.9062,
+      "step": 3409
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.748276555528981e-06,
+      "loss": 0.8881,
+      "step": 3410
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.742563949504571e-06,
+      "loss": 1.1004,
+      "step": 3411
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.73685371358544e-06,
+      "loss": 1.0467,
+      "step": 3412
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.731145850345826e-06,
+      "loss": 1.0586,
+      "step": 3413
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.725440362358877e-06,
+      "loss": 1.0071,
+      "step": 3414
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.719737252196698e-06,
+      "loss": 1.171,
+      "step": 3415
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.714036522430302e-06,
+      "loss": 1.0819,
+      "step": 3416
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.708338175629632e-06,
+      "loss": 1.0204,
+      "step": 3417
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.70264221436357e-06,
+      "loss": 1.1554,
+      "step": 3418
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.696948641199906e-06,
+      "loss": 0.9106,
+      "step": 3419
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.691257458705361e-06,
+      "loss": 0.9944,
+      "step": 3420
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.6855686694455845e-06,
+      "loss": 0.8232,
+      "step": 3421
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.679882275985127e-06,
+      "loss": 1.0215,
+      "step": 3422
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.674198280887479e-06,
+      "loss": 0.8846,
+      "step": 3423
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.6685166867150525e-06,
+      "loss": 0.8854,
+      "step": 3424
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.662837496029149e-06,
+      "loss": 0.9478,
+      "step": 3425
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.657160711390018e-06,
+      "loss": 0.8529,
+      "step": 3426
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.65148633535681e-06,
+      "loss": 0.9744,
+      "step": 3427
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.645814370487583e-06,
+      "loss": 1.0778,
+      "step": 3428
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.640144819339326e-06,
+      "loss": 0.8583,
+      "step": 3429
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.634477684467927e-06,
+      "loss": 0.9765,
+      "step": 3430
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.628812968428181e-06,
+      "loss": 1.2319,
+      "step": 3431
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.62315067377381e-06,
+      "loss": 0.9552,
+      "step": 3432
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.617490803057427e-06,
+      "loss": 1.0201,
+      "step": 3433
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.61183335883056e-06,
+      "loss": 1.1325,
+      "step": 3434
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.606178343643639e-06,
+      "loss": 1.1578,
+      "step": 3435
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.60052576004601e-06,
+      "loss": 0.9037,
+      "step": 3436
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.594875610585908e-06,
+      "loss": 1.1286,
+      "step": 3437
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.589227897810477e-06,
+      "loss": 1.089,
+      "step": 3438
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.583582624265768e-06,
+      "loss": 0.8419,
+      "step": 3439
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.577939792496725e-06,
+      "loss": 1.0841,
+      "step": 3440
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.5722994050471904e-06,
+      "loss": 1.0853,
+      "step": 3441
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.5666614644599124e-06,
+      "loss": 1.2503,
+      "step": 3442
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.561025973276532e-06,
+      "loss": 0.9831,
+      "step": 3443
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.555392934037579e-06,
+      "loss": 1.0397,
+      "step": 3444
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.549762349282496e-06,
+      "loss": 1.0993,
+      "step": 3445
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.544134221549592e-06,
+      "loss": 0.9428,
+      "step": 3446
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.538508553376098e-06,
+      "loss": 0.8436,
+      "step": 3447
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.532885347298114e-06,
+      "loss": 0.9791,
+      "step": 3448
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.527264605850637e-06,
+      "loss": 1.0938,
+      "step": 3449
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.521646331567562e-06,
+      "loss": 0.8554,
+      "step": 3450
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.5160305269816574e-06,
+      "loss": 1.1532,
+      "step": 3451
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.510417194624583e-06,
+      "loss": 0.9703,
+      "step": 3452
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.504806337026898e-06,
+      "loss": 1.0543,
+      "step": 3453
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.499197956718017e-06,
+      "loss": 0.9302,
+      "step": 3454
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.49359205622626e-06,
+      "loss": 0.993,
+      "step": 3455
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.487988638078835e-06,
+      "loss": 1.0467,
+      "step": 3456
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.482387704801803e-06,
+      "loss": 1.4069,
+      "step": 3457
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.476789258920131e-06,
+      "loss": 0.9442,
+      "step": 3458
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.471193302957654e-06,
+      "loss": 0.9281,
+      "step": 3459
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.465599839437079e-06,
+      "loss": 1.0676,
+      "step": 3460
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.460008870880005e-06,
+      "loss": 0.9784,
+      "step": 3461
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.454420399806894e-06,
+      "loss": 1.2023,
+      "step": 3462
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.448834428737079e-06,
+      "loss": 1.0646,
+      "step": 3463
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.443250960188782e-06,
+      "loss": 1.0609,
+      "step": 3464
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.4376699966790814e-06,
+      "loss": 1.0969,
+      "step": 3465
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.43209154072393e-06,
+      "loss": 1.1207,
+      "step": 3466
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.426515594838158e-06,
+      "loss": 1.01,
+      "step": 3467
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.4209421615354555e-06,
+      "loss": 1.0166,
+      "step": 3468
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.415371243328382e-06,
+      "loss": 1.0403,
+      "step": 3469
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.40980284272836e-06,
+      "loss": 0.8547,
+      "step": 3470
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.40423696224569e-06,
+      "loss": 1.0949,
+      "step": 3471
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.398673604389519e-06,
+      "loss": 1.0402,
+      "step": 3472
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.393112771667865e-06,
+      "loss": 1.0928,
+      "step": 3473
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.387554466587615e-06,
+      "loss": 1.2161,
+      "step": 3474
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.381998691654503e-06,
+      "loss": 1.0445,
+      "step": 3475
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.376445449373131e-06,
+      "loss": 0.9656,
+      "step": 3476
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.370894742246955e-06,
+      "loss": 0.9879,
+      "step": 3477
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.365346572778286e-06,
+      "loss": 0.9168,
+      "step": 3478
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.359800943468303e-06,
+      "loss": 0.9168,
+      "step": 3479
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.354257856817028e-06,
+      "loss": 1.2126,
+      "step": 3480
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.348717315323337e-06,
+      "loss": 0.9531,
+      "step": 3481
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.343179321484968e-06,
+      "loss": 1.0343,
+      "step": 3482
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.337643877798503e-06,
+      "loss": 1.0333,
+      "step": 3483
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.332110986759371e-06,
+      "loss": 0.9899,
+      "step": 3484
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.326580650861862e-06,
+      "loss": 0.9087,
+      "step": 3485
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.321052872599105e-06,
+      "loss": 0.921,
+      "step": 3486
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.315527654463074e-06,
+      "loss": 0.9823,
+      "step": 3487
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.3100049989446035e-06,
+      "loss": 1.129,
+      "step": 3488
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.3044849085333495e-06,
+      "loss": 0.7683,
+      "step": 3489
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.2989673857178295e-06,
+      "loss": 0.921,
+      "step": 3490
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.2934524329854075e-06,
+      "loss": 1.1359,
+      "step": 3491
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.287940052822265e-06,
+      "loss": 1.0255,
+      "step": 3492
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.282430247713449e-06,
+      "loss": 1.0368,
+      "step": 3493
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.276923020142832e-06,
+      "loss": 0.9948,
+      "step": 3494
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.271418372593124e-06,
+      "loss": 1.1219,
+      "step": 3495
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.265916307545884e-06,
+      "loss": 0.9041,
+      "step": 3496
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.260416827481494e-06,
+      "loss": 1.0503,
+      "step": 3497
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.254919934879171e-06,
+      "loss": 1.1054,
+      "step": 3498
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.249425632216978e-06,
+      "loss": 0.9307,
+      "step": 3499
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.243933921971798e-06,
+      "loss": 1.0259,
+      "step": 3500
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.23844480661935e-06,
+      "loss": 1.1614,
+      "step": 3501
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.232958288634178e-06,
+      "loss": 1.2352,
+      "step": 3502
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.227474370489667e-06,
+      "loss": 1.0372,
+      "step": 3503
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.22199305465802e-06,
+      "loss": 0.8876,
+      "step": 3504
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.21651434361027e-06,
+      "loss": 1.0004,
+      "step": 3505
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.21103823981627e-06,
+      "loss": 1.1055,
+      "step": 3506
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.205564745744712e-06,
+      "loss": 1.0133,
+      "step": 3507
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.200093863863096e-06,
+      "loss": 1.1009,
+      "step": 3508
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.19462559663775e-06,
+      "loss": 1.0201,
+      "step": 3509
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.1891599465338305e-06,
+      "loss": 0.9374,
+      "step": 3510
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.183696916015304e-06,
+      "loss": 1.0304,
+      "step": 3511
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.1782365075449595e-06,
+      "loss": 0.9678,
+      "step": 3512
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.172778723584401e-06,
+      "loss": 0.8689,
+      "step": 3513
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.167323566594061e-06,
+      "loss": 1.1213,
+      "step": 3514
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.161871039033173e-06,
+      "loss": 0.8753,
+      "step": 3515
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.15642114335979e-06,
+      "loss": 0.9494,
+      "step": 3516
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.150973882030787e-06,
+      "loss": 0.9477,
+      "step": 3517
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.14552925750184e-06,
+      "loss": 0.9609,
+      "step": 3518
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.140087272227438e-06,
+      "loss": 0.7795,
+      "step": 3519
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.13464792866089e-06,
+      "loss": 1.2401,
+      "step": 3520
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.129211229254303e-06,
+      "loss": 1.1841,
+      "step": 3521
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.123777176458592e-06,
+      "loss": 0.982,
+      "step": 3522
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.118345772723495e-06,
+      "loss": 1.0835,
+      "step": 3523
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.11291702049753e-06,
+      "loss": 1.0727,
+      "step": 3524
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.107490922228038e-06,
+      "loss": 1.0432,
+      "step": 3525
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.102067480361167e-06,
+      "loss": 0.9268,
+      "step": 3526
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.096646697341845e-06,
+      "loss": 1.0402,
+      "step": 3527
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.091228575613828e-06,
+      "loss": 1.0793,
+      "step": 3528
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.0858131176196545e-06,
+      "loss": 1.0952,
+      "step": 3529
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.080400325800666e-06,
+      "loss": 1.0426,
+      "step": 3530
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.0749902025970115e-06,
+      "loss": 1.2086,
+      "step": 3531
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.069582750447617e-06,
+      "loss": 0.8669,
+      "step": 3532
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.064177971790223e-06,
+      "loss": 1.0424,
+      "step": 3533
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.058775869061366e-06,
+      "loss": 0.8696,
+      "step": 3534
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.053376444696356e-06,
+      "loss": 1.0271,
+      "step": 3535
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.047979701129314e-06,
+      "loss": 1.1018,
+      "step": 3536
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.042585640793148e-06,
+      "loss": 1.1339,
+      "step": 3537
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.037194266119546e-06,
+      "loss": 1.0627,
+      "step": 3538
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.031805579539006e-06,
+      "loss": 0.9147,
+      "step": 3539
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.0264195834807984e-06,
+      "loss": 1.2464,
+      "step": 3540
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.021036280372979e-06,
+      "loss": 0.9122,
+      "step": 3541
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.015655672642402e-06,
+      "loss": 0.8989,
+      "step": 3542
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.010277762714698e-06,
+      "loss": 0.9092,
+      "step": 3543
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.004902553014278e-06,
+      "loss": 0.984,
+      "step": 3544
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.9995300459643504e-06,
+      "loss": 0.7907,
+      "step": 3545
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.994160243986889e-06,
+      "loss": 1.0913,
+      "step": 3546
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.988793149502657e-06,
+      "loss": 1.0311,
+      "step": 3547
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.9834287649311895e-06,
+      "loss": 0.9589,
+      "step": 3548
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.978067092690815e-06,
+      "loss": 0.8385,
+      "step": 3549
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.972708135198623e-06,
+      "loss": 0.9262,
+      "step": 3550
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.967351894870484e-06,
+      "loss": 1.0565,
+      "step": 3551
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.961998374121051e-06,
+      "loss": 1.2417,
+      "step": 3552
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.956647575363744e-06,
+      "loss": 0.9275,
+      "step": 3553
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.95129950101075e-06,
+      "loss": 1.1528,
+      "step": 3554
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.9459541534730484e-06,
+      "loss": 1.2051,
+      "step": 3555
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.940611535160361e-06,
+      "loss": 0.9263,
+      "step": 3556
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.935271648481202e-06,
+      "loss": 1.0473,
+      "step": 3557
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.929934495842851e-06,
+      "loss": 1.105,
+      "step": 3558
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.924600079651338e-06,
+      "loss": 1.1453,
+      "step": 3559
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.919268402311481e-06,
+      "loss": 1.0335,
+      "step": 3560
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.913939466226853e-06,
+      "loss": 1.1741,
+      "step": 3561
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.908613273799786e-06,
+      "loss": 0.941,
+      "step": 3562
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.9032898274313905e-06,
+      "loss": 1.0993,
+      "step": 3563
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.897969129521524e-06,
+      "loss": 1.1025,
+      "step": 3564
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.892651182468809e-06,
+      "loss": 1.0065,
+      "step": 3565
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.887335988670641e-06,
+      "loss": 1.0506,
+      "step": 3566
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.8820235505231465e-06,
+      "loss": 0.9392,
+      "step": 3567
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.876713870421236e-06,
+      "loss": 0.9233,
+      "step": 3568
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.871406950758571e-06,
+      "loss": 0.9638,
+      "step": 3569
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.866102793927554e-06,
+      "loss": 1.032,
+      "step": 3570
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.8608014023193595e-06,
+      "loss": 0.9941,
+      "step": 3571
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.855502778323908e-06,
+      "loss": 1.1745,
+      "step": 3572
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.850206924329867e-06,
+      "loss": 0.9797,
+      "step": 3573
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.844913842724669e-06,
+      "loss": 0.8986,
+      "step": 3574
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.8396235358944846e-06,
+      "loss": 0.8444,
+      "step": 3575
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.834336006224235e-06,
+      "loss": 1.0726,
+      "step": 3576
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.829051256097598e-06,
+      "loss": 0.8891,
+      "step": 3577
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.823769287896989e-06,
+      "loss": 0.9674,
+      "step": 3578
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.818490104003572e-06,
+      "loss": 1.1041,
+      "step": 3579
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.813213706797255e-06,
+      "loss": 0.8335,
+      "step": 3580
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.807940098656696e-06,
+      "loss": 0.9977,
+      "step": 3581
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.8026692819592893e-06,
+      "loss": 0.8463,
+      "step": 3582
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.7974012590811672e-06,
+      "loss": 0.9032,
+      "step": 3583
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.7921360323972143e-06,
+      "loss": 1.0629,
+      "step": 3584
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.786873604281046e-06,
+      "loss": 0.9186,
+      "step": 3585
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.781613977105013e-06,
+      "loss": 0.8059,
+      "step": 3586
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.7763571532402176e-06,
+      "loss": 1.1629,
+      "step": 3587
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.771103135056483e-06,
+      "loss": 0.9722,
+      "step": 3588
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.7658519249223756e-06,
+      "loss": 0.9517,
+      "step": 3589
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.7606035252051933e-06,
+      "loss": 1.0078,
+      "step": 3590
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.7553579382709647e-06,
+      "loss": 0.8349,
+      "step": 3591
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.7501151664844593e-06,
+      "loss": 1.0695,
+      "step": 3592
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.7448752122091682e-06,
+      "loss": 0.9156,
+      "step": 3593
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.7396380778073127e-06,
+      "loss": 0.9054,
+      "step": 3594
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.7344037656398535e-06,
+      "loss": 0.8722,
+      "step": 3595
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.7291722780664665e-06,
+      "loss": 1.0914,
+      "step": 3596
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.723943617445557e-06,
+      "loss": 1.0151,
+      "step": 3597
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.718717786134267e-06,
+      "loss": 1.1105,
+      "step": 3598
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.713494786488442e-06,
+      "loss": 1.6535,
+      "step": 3599
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.7082746208626675e-06,
+      "loss": 0.8861,
+      "step": 3600
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.703057291610257e-06,
+      "loss": 1.0971,
+      "step": 3601
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6978428010832177e-06,
+      "loss": 1.0357,
+      "step": 3602
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.692631151632309e-06,
+      "loss": 1.0465,
+      "step": 3603
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.687422345606988e-06,
+      "loss": 1.0766,
+      "step": 3604
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.682216385355436e-06,
+      "loss": 0.9432,
+      "step": 3605
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.677013273224559e-06,
+      "loss": 0.7886,
+      "step": 3606
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.671813011559967e-06,
+      "loss": 0.6898,
+      "step": 3607
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.6666156027059895e-06,
+      "loss": 1.0338,
+      "step": 3608
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.661421049005678e-06,
+      "loss": 0.8382,
+      "step": 3609
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.656229352800784e-06,
+      "loss": 1.2502,
+      "step": 3610
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.6510405164317753e-06,
+      "loss": 0.9461,
+      "step": 3611
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.6458545422378377e-06,
+      "loss": 1.2323,
+      "step": 3612
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.640671432556857e-06,
+      "loss": 0.9841,
+      "step": 3613
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.635491189725432e-06,
+      "loss": 1.1001,
+      "step": 3614
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.630313816078864e-06,
+      "loss": 0.9071,
+      "step": 3615
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.6251393139511735e-06,
+      "loss": 0.9846,
+      "step": 3616
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.6199676856750756e-06,
+      "loss": 1.0945,
+      "step": 3617
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.61479893358199e-06,
+      "loss": 1.1434,
+      "step": 3618
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.609633060002041e-06,
+      "loss": 1.0674,
+      "step": 3619
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.6044700672640633e-06,
+      "loss": 0.9817,
+      "step": 3620
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5993099576955813e-06,
+      "loss": 0.9686,
+      "step": 3621
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.5941527336228223e-06,
+      "loss": 1.082,
+      "step": 3622
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.588998397370722e-06,
+      "loss": 0.904,
+      "step": 3623
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.5838469512629026e-06,
+      "loss": 0.9767,
+      "step": 3624
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.5786983976216893e-06,
+      "loss": 1.1961,
+      "step": 3625
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.573552738768099e-06,
+      "loss": 1.0585,
+      "step": 3626
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.568409977021852e-06,
+      "loss": 0.9152,
+      "step": 3627
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.5632701147013548e-06,
+      "loss": 1.0969,
+      "step": 3628
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.558133154123705e-06,
+      "loss": 0.984,
+      "step": 3629
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.5529990976047046e-06,
+      "loss": 1.0255,
+      "step": 3630
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.547867947458834e-06,
+      "loss": 0.896,
+      "step": 3631
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.542739705999263e-06,
+      "loss": 1.045,
+      "step": 3632
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.5376143755378667e-06,
+      "loss": 0.9164,
+      "step": 3633
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.5324919583851814e-06,
+      "loss": 1.1281,
+      "step": 3634
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.5273724568504506e-06,
+      "loss": 1.1079,
+      "step": 3635
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.5222558732416056e-06,
+      "loss": 1.1072,
+      "step": 3636
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.5171422098652385e-06,
+      "loss": 0.9112,
+      "step": 3637
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.5120314690266533e-06,
+      "loss": 0.9177,
+      "step": 3638
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.5069236530298177e-06,
+      "loss": 1.1832,
+      "step": 3639
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.501818764177384e-06,
+      "loss": 1.0136,
+      "step": 3640
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.4967168047706944e-06,
+      "loss": 1.1482,
+      "step": 3641
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.4916177771097617e-06,
+      "loss": 1.1756,
+      "step": 3642
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.4865216834932734e-06,
+      "loss": 0.9993,
+      "step": 3643
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.481428526218612e-06,
+      "loss": 1.0682,
+      "step": 3644
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.4763383075818102e-06,
+      "loss": 1.0667,
+      "step": 3645
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.471251029877597e-06,
+      "loss": 1.1463,
+      "step": 3646
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.4661666953993743e-06,
+      "loss": 1.1522,
+      "step": 3647
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.4610853064391993e-06,
+      "loss": 0.9347,
+      "step": 3648
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.456006865287822e-06,
+      "loss": 0.9445,
+      "step": 3649
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.4509313742346526e-06,
+      "loss": 0.8468,
+      "step": 3650
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.44585883556777e-06,
+      "loss": 0.9702,
+      "step": 3651
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.4407892515739337e-06,
+      "loss": 0.92,
+      "step": 3652
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.4357226245385587e-06,
+      "loss": 0.9829,
+      "step": 3653
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.430658956745728e-06,
+      "loss": 1.0523,
+      "step": 3654
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.4255982504782027e-06,
+      "loss": 1.0755,
+      "step": 3655
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.420540508017396e-06,
+      "loss": 0.9693,
+      "step": 3656
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.4154857316433886e-06,
+      "loss": 1.0413,
+      "step": 3657
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.4104339236349227e-06,
+      "loss": 1.1437,
+      "step": 3658
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.4053850862694094e-06,
+      "loss": 1.0123,
+      "step": 3659
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.400339221822914e-06,
+      "loss": 1.0819,
+      "step": 3660
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.3952963325701584e-06,
+      "loss": 1.2525,
+      "step": 3661
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.3902564207845345e-06,
+      "loss": 1.0336,
+      "step": 3662
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.385219488738084e-06,
+      "loss": 1.1898,
+      "step": 3663
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.3801855387015026e-06,
+      "loss": 0.9181,
+      "step": 3664
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.3751545729441515e-06,
+      "loss": 0.9668,
+      "step": 3665
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.3701265937340387e-06,
+      "loss": 1.0466,
+      "step": 3666
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.365101603337825e-06,
+      "loss": 0.984,
+      "step": 3667
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.360079604020835e-06,
+      "loss": 1.0755,
+      "step": 3668
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.355060598047024e-06,
+      "loss": 1.0353,
+      "step": 3669
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.350044587679019e-06,
+      "loss": 1.0879,
+      "step": 3670
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.3450315751780925e-06,
+      "loss": 1.0612,
+      "step": 3671
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.34002156280415e-06,
+      "loss": 1.0945,
+      "step": 3672
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.335014552815764e-06,
+      "loss": 1.0689,
+      "step": 3673
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.3300105474701415e-06,
+      "loss": 0.7961,
+      "step": 3674
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.3250095490231373e-06,
+      "loss": 1.0013,
+      "step": 3675
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.3200115597292603e-06,
+      "loss": 1.1697,
+      "step": 3676
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.3150165818416424e-06,
+      "loss": 0.9957,
+      "step": 3677
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.3100246176120765e-06,
+      "loss": 0.8069,
+      "step": 3678
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.3050356692909967e-06,
+      "loss": 0.9935,
+      "step": 3679
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.3000497391274576e-06,
+      "loss": 1.011,
+      "step": 3680
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.295066829369178e-06,
+      "loss": 0.8351,
+      "step": 3681
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2900869422624992e-06,
+      "loss": 1.3015,
+      "step": 3682
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.285110080052403e-06,
+      "loss": 0.8962,
+      "step": 3683
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2801362449825145e-06,
+      "loss": 0.9538,
+      "step": 3684
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2751654392950873e-06,
+      "loss": 1.0419,
+      "step": 3685
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.270197665231004e-06,
+      "loss": 1.0689,
+      "step": 3686
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2652329250297978e-06,
+      "loss": 0.951,
+      "step": 3687
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.260271220929617e-06,
+      "loss": 1.1354,
+      "step": 3688
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.2553125551672482e-06,
+      "loss": 0.9208,
+      "step": 3689
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.250356929978111e-06,
+      "loss": 1.0886,
+      "step": 3690
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.2454043475962503e-06,
+      "loss": 1.085,
+      "step": 3691
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.240454810254341e-06,
+      "loss": 1.0531,
+      "step": 3692
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.2355083201836777e-06,
+      "loss": 1.1485,
+      "step": 3693
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.2305648796141986e-06,
+      "loss": 0.9192,
+      "step": 3694
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.2256244907744517e-06,
+      "loss": 1.0314,
+      "step": 3695
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.2206871558916098e-06,
+      "loss": 1.101,
+      "step": 3696
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.2157528771914814e-06,
+      "loss": 1.1065,
+      "step": 3697
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.210821656898485e-06,
+      "loss": 1.0542,
+      "step": 3698
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.2058934972356625e-06,
+      "loss": 1.1043,
+      "step": 3699
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.2009684004246876e-06,
+      "loss": 0.9377,
+      "step": 3700
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.196046368685829e-06,
+      "loss": 0.8958,
+      "step": 3701
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.191127404237999e-06,
+      "loss": 1.1758,
+      "step": 3702
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1862115092987144e-06,
+      "loss": 0.9829,
+      "step": 3703
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.1812986860841053e-06,
+      "loss": 1.0692,
+      "step": 3704
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.1763889368089307e-06,
+      "loss": 1.0252,
+      "step": 3705
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.1714822636865496e-06,
+      "loss": 1.0003,
+      "step": 3706
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.166578668928937e-06,
+      "loss": 0.9379,
+      "step": 3707
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.1616781547466904e-06,
+      "loss": 1.0875,
+      "step": 3708
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.156780723349007e-06,
+      "loss": 0.7864,
+      "step": 3709
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.151886376943696e-06,
+      "loss": 0.992,
+      "step": 3710
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.1469951177371873e-06,
+      "loss": 0.9195,
+      "step": 3711
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.1421069479344966e-06,
+      "loss": 0.8308,
+      "step": 3712
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.1372218697392677e-06,
+      "loss": 0.9725,
+      "step": 3713
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.132339885353749e-06,
+      "loss": 1.0345,
+      "step": 3714
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.127460996978775e-06,
+      "loss": 1.0639,
+      "step": 3715
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.1225852068138097e-06,
+      "loss": 1.0545,
+      "step": 3716
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.117712517056902e-06,
+      "loss": 1.1479,
+      "step": 3717
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.112842929904709e-06,
+      "loss": 1.144,
+      "step": 3718
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.1079764475524967e-06,
+      "loss": 0.7748,
+      "step": 3719
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.1031130721941183e-06,
+      "loss": 1.1182,
+      "step": 3720
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.098252806022032e-06,
+      "loss": 0.9762,
+      "step": 3721
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.0933956512273e-06,
+      "loss": 0.9502,
+      "step": 3722
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.0885416099995736e-06,
+      "loss": 1.1481,
+      "step": 3723
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.0836906845271043e-06,
+      "loss": 0.9445,
+      "step": 3724
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.0788428769967348e-06,
+      "loss": 0.8324,
+      "step": 3725
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.0739981895939118e-06,
+      "loss": 0.8768,
+      "step": 3726
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.069156624502666e-06,
+      "loss": 1.0679,
+      "step": 3727
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.0643181839056214e-06,
+      "loss": 1.0548,
+      "step": 3728
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.0594828699840017e-06,
+      "loss": 1.0784,
+      "step": 3729
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.054650684917613e-06,
+      "loss": 0.9586,
+      "step": 3730
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.0498216308848516e-06,
+      "loss": 1.0364,
+      "step": 3731
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.0449957100627016e-06,
+      "loss": 1.1083,
+      "step": 3732
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.040172924626743e-06,
+      "loss": 0.7586,
+      "step": 3733
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.035353276751134e-06,
+      "loss": 0.8131,
+      "step": 3734
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.0305367686086186e-06,
+      "loss": 1.0471,
+      "step": 3735
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.025723402370525e-06,
+      "loss": 0.9811,
+      "step": 3736
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.020913180206775e-06,
+      "loss": 1.0382,
+      "step": 3737
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.0161061042858596e-06,
+      "loss": 0.8843,
+      "step": 3738
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.011302176774856e-06,
+      "loss": 1.1373,
+      "step": 3739
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.0065013998394287e-06,
+      "loss": 0.8857,
+      "step": 3740
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.001703775643814e-06,
+      "loss": 1.0259,
+      "step": 3741
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.996909306350826e-06,
+      "loss": 1.0427,
+      "step": 3742
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.992117994121866e-06,
+      "loss": 0.9261,
+      "step": 3743
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9873298411169015e-06,
+      "loss": 1.0925,
+      "step": 3744
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.98254484949448e-06,
+      "loss": 1.2307,
+      "step": 3745
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9777630214117305e-06,
+      "loss": 1.0072,
+      "step": 3746
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.972984359024339e-06,
+      "loss": 0.8088,
+      "step": 3747
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9682088644865838e-06,
+      "loss": 0.8858,
+      "step": 3748
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.963436539951301e-06,
+      "loss": 0.9585,
+      "step": 3749
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9586673875699025e-06,
+      "loss": 0.8225,
+      "step": 3750
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9539014094923745e-06,
+      "loss": 1.1588,
+      "step": 3751
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.949138607867266e-06,
+      "loss": 1.1331,
+      "step": 3752
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9443789848416917e-06,
+      "loss": 0.8911,
+      "step": 3753
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9396225425613457e-06,
+      "loss": 0.8982,
+      "step": 3754
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.934869283170476e-06,
+      "loss": 0.903,
+      "step": 3755
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9301192088118967e-06,
+      "loss": 1.2268,
+      "step": 3756
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9253723216270003e-06,
+      "loss": 1.2382,
+      "step": 3757
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.920628623755717e-06,
+      "loss": 1.0442,
+      "step": 3758
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9158881173365638e-06,
+      "loss": 1.1556,
+      "step": 3759
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.911150804506606e-06,
+      "loss": 0.9318,
+      "step": 3760
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.906416687401469e-06,
+      "loss": 0.8563,
+      "step": 3761
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.901685768155347e-06,
+      "loss": 0.8421,
+      "step": 3762
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.896958048900982e-06,
+      "loss": 1.025,
+      "step": 3763
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.892233531769675e-06,
+      "loss": 1.0014,
+      "step": 3764
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.887512218891293e-06,
+      "loss": 1.049,
+      "step": 3765
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8827941123942472e-06,
+      "loss": 1.0033,
+      "step": 3766
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.878079214405506e-06,
+      "loss": 1.0565,
+      "step": 3767
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.873367527050598e-06,
+      "loss": 0.9989,
+      "step": 3768
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8686590524535974e-06,
+      "loss": 0.9536,
+      "step": 3769
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.863953792737131e-06,
+      "loss": 1.0819,
+      "step": 3770
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.8592517500223748e-06,
+      "loss": 0.9813,
+      "step": 3771
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.854552926429064e-06,
+      "loss": 0.9281,
+      "step": 3772
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.849857324075472e-06,
+      "loss": 1.2957,
+      "step": 3773
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.84516494507842e-06,
+      "loss": 0.9512,
+      "step": 3774
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.8404757915532875e-06,
+      "loss": 1.0178,
+      "step": 3775
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.8357898656139883e-06,
+      "loss": 0.8709,
+      "step": 3776
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.831107169372982e-06,
+      "loss": 1.0216,
+      "step": 3777
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.8264277049412847e-06,
+      "loss": 0.8214,
+      "step": 3778
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.8217514744284345e-06,
+      "loss": 1.0333,
+      "step": 3779
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.8170784799425288e-06,
+      "loss": 1.1535,
+      "step": 3780
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.8124087235902074e-06,
+      "loss": 0.9592,
+      "step": 3781
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.8077422074766302e-06,
+      "loss": 1.0066,
+      "step": 3782
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.803078933705521e-06,
+      "loss": 0.8733,
+      "step": 3783
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7984189043791266e-06,
+      "loss": 0.9672,
+      "step": 3784
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7937621215982314e-06,
+      "loss": 1.0302,
+      "step": 3785
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.789108587462167e-06,
+      "loss": 0.6601,
+      "step": 3786
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.784458304068791e-06,
+      "loss": 0.9557,
+      "step": 3787
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.7798112735144943e-06,
+      "loss": 0.9167,
+      "step": 3788
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.7751674978942147e-06,
+      "loss": 1.0835,
+      "step": 3789
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.7705269793014e-06,
+      "loss": 1.0053,
+      "step": 3790
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.765889719828051e-06,
+      "loss": 1.1799,
+      "step": 3791
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.761255721564696e-06,
+      "loss": 1.0558,
+      "step": 3792
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.756624986600376e-06,
+      "loss": 1.0249,
+      "step": 3793
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.751997517022683e-06,
+      "loss": 1.0927,
+      "step": 3794
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.747373314917723e-06,
+      "loss": 1.2521,
+      "step": 3795
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.7427523823701298e-06,
+      "loss": 0.8255,
+      "step": 3796
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.7381347214630726e-06,
+      "loss": 1.1121,
+      "step": 3797
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.733520334278238e-06,
+      "loss": 0.9841,
+      "step": 3798
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.7289092228958348e-06,
+      "loss": 0.9656,
+      "step": 3799
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.7243013893946036e-06,
+      "loss": 1.0947,
+      "step": 3800
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.7196968358517992e-06,
+      "loss": 0.9656,
+      "step": 3801
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.715095564343202e-06,
+      "loss": 1.0124,
+      "step": 3802
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.7104975769431087e-06,
+      "loss": 1.0854,
+      "step": 3803
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.7059028757243434e-06,
+      "loss": 1.044,
+      "step": 3804
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.7013114627582416e-06,
+      "loss": 0.9879,
+      "step": 3805
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.696723340114655e-06,
+      "loss": 1.0121,
+      "step": 3806
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.692138509861962e-06,
+      "loss": 0.793,
+      "step": 3807
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.687556974067048e-06,
+      "loss": 1.0602,
+      "step": 3808
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.682978734795312e-06,
+      "loss": 1.1511,
+      "step": 3809
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6784037941106768e-06,
+      "loss": 1.1231,
+      "step": 3810
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6738321540755673e-06,
+      "loss": 1.1204,
+      "step": 3811
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6692638167509244e-06,
+      "loss": 1.1272,
+      "step": 3812
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.664698784196209e-06,
+      "loss": 0.8995,
+      "step": 3813
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6601370584693707e-06,
+      "loss": 0.9336,
+      "step": 3814
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6555786416268924e-06,
+      "loss": 1.0378,
+      "step": 3815
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.651023535723749e-06,
+      "loss": 0.9132,
+      "step": 3816
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.646471742813428e-06,
+      "loss": 1.1491,
+      "step": 3817
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6419232649479275e-06,
+      "loss": 1.0529,
+      "step": 3818
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.6373781041777458e-06,
+      "loss": 1.1695,
+      "step": 3819
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.6328362625518823e-06,
+      "loss": 1.0844,
+      "step": 3820
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.628297742117857e-06,
+      "loss": 0.7919,
+      "step": 3821
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.6237625449216666e-06,
+      "loss": 1.253,
+      "step": 3822
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.6192306730078287e-06,
+      "loss": 0.9696,
+      "step": 3823
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.6147021284193662e-06,
+      "loss": 1.1289,
+      "step": 3824
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.6101769131977784e-06,
+      "loss": 1.3368,
+      "step": 3825
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.6056550293830886e-06,
+      "loss": 0.978,
+      "step": 3826
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.6011364790138027e-06,
+      "loss": 0.9171,
+      "step": 3827
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.596621264126928e-06,
+      "loss": 0.896,
+      "step": 3828
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.592109386757974e-06,
+      "loss": 1.0415,
+      "step": 3829
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5876008489409364e-06,
+      "loss": 0.9463,
+      "step": 3830
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5830956527083082e-06,
+      "loss": 1.1791,
+      "step": 3831
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5785938000910837e-06,
+      "loss": 1.1109,
+      "step": 3832
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.574095293118739e-06,
+      "loss": 1.0679,
+      "step": 3833
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5696001338192446e-06,
+      "loss": 0.9588,
+      "step": 3834
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.565108324219069e-06,
+      "loss": 1.0011,
+      "step": 3835
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.5606198663431626e-06,
+      "loss": 0.9563,
+      "step": 3836
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.5561347622149677e-06,
+      "loss": 1.331,
+      "step": 3837
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.5516530138564123e-06,
+      "loss": 1.2691,
+      "step": 3838
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.547174623287918e-06,
+      "loss": 0.9194,
+      "step": 3839
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.5426995925283872e-06,
+      "loss": 1.0811,
+      "step": 3840
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.538227923595208e-06,
+      "loss": 0.9282,
+      "step": 3841
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.533759618504251e-06,
+      "loss": 0.8187,
+      "step": 3842
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.529294679269879e-06,
+      "loss": 0.9693,
+      "step": 3843
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.5248331079049305e-06,
+      "loss": 0.9855,
+      "step": 3844
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.520374906420725e-06,
+      "loss": 0.8643,
+      "step": 3845
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.515920076827062e-06,
+      "loss": 1.1674,
+      "step": 3846
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.5114686211322303e-06,
+      "loss": 0.8284,
+      "step": 3847
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.5070205413429884e-06,
+      "loss": 0.8287,
+      "step": 3848
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.5025758394645714e-06,
+      "loss": 1.0208,
+      "step": 3849
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4981345175007032e-06,
+      "loss": 0.8322,
+      "step": 3850
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4936965774535727e-06,
+      "loss": 1.0102,
+      "step": 3851
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.489262021323845e-06,
+      "loss": 0.8587,
+      "step": 3852
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.4848308511106688e-06,
+      "loss": 1.0065,
+      "step": 3853
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.480403068811659e-06,
+      "loss": 1.2534,
+      "step": 3854
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.475978676422899e-06,
+      "loss": 1.0187,
+      "step": 3855
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.4715576759389594e-06,
+      "loss": 0.9398,
+      "step": 3856
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.4671400693528613e-06,
+      "loss": 0.9319,
+      "step": 3857
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.462725858656111e-06,
+      "loss": 1.0009,
+      "step": 3858
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.4583150458386864e-06,
+      "loss": 0.9536,
+      "step": 3859
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.453907632889012e-06,
+      "loss": 1.1335,
+      "step": 3860
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.449503621794006e-06,
+      "loss": 1.1438,
+      "step": 3861
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.4451030145390376e-06,
+      "loss": 0.8636,
+      "step": 3862
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.44070581310794e-06,
+      "loss": 1.1725,
+      "step": 3863
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.4363120194830235e-06,
+      "loss": 0.8697,
+      "step": 3864
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.4319216356450514e-06,
+      "loss": 1.0413,
+      "step": 3865
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.4275346635732513e-06,
+      "loss": 1.1494,
+      "step": 3866
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.4231511052453185e-06,
+      "loss": 1.1792,
+      "step": 3867
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.4187709626374032e-06,
+      "loss": 1.2548,
+      "step": 3868
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.414394237724118e-06,
+      "loss": 0.9455,
+      "step": 3869
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.410020932478535e-06,
+      "loss": 1.1615,
+      "step": 3870
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.4056510488721805e-06,
+      "loss": 0.8031,
+      "step": 3871
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.401284588875049e-06,
+      "loss": 1.1338,
+      "step": 3872
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3969215544555825e-06,
+      "loss": 0.8838,
+      "step": 3873
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.392561947580676e-06,
+      "loss": 0.9179,
+      "step": 3874
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3882057702156913e-06,
+      "loss": 1.054,
+      "step": 3875
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.383853024324434e-06,
+      "loss": 1.0566,
+      "step": 3876
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3795037118691632e-06,
+      "loss": 0.9558,
+      "step": 3877
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3751578348106e-06,
+      "loss": 1.0797,
+      "step": 3878
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.370815395107904e-06,
+      "loss": 1.1332,
+      "step": 3879
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.366476394718693e-06,
+      "loss": 0.8737,
+      "step": 3880
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3621408355990283e-06,
+      "loss": 1.0462,
+      "step": 3881
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3578087197034296e-06,
+      "loss": 0.996,
+      "step": 3882
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3534800489848563e-06,
+      "loss": 1.0176,
+      "step": 3883
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3491548253947138e-06,
+      "loss": 1.1201,
+      "step": 3884
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.3448330508828622e-06,
+      "loss": 1.0914,
+      "step": 3885
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.340514727397597e-06,
+      "loss": 0.9287,
+      "step": 3886
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.3361998568856615e-06,
+      "loss": 1.131,
+      "step": 3887
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.331888441292247e-06,
+      "loss": 0.987,
+      "step": 3888
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.3275804825609815e-06,
+      "loss": 1.0587,
+      "step": 3889
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.3232759826339323e-06,
+      "loss": 1.0721,
+      "step": 3890
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.3189749434516206e-06,
+      "loss": 0.9582,
+      "step": 3891
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.3146773669529865e-06,
+      "loss": 1.1893,
+      "step": 3892
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.3103832550754303e-06,
+      "loss": 1.144,
+      "step": 3893
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.3060926097547762e-06,
+      "loss": 0.9633,
+      "step": 3894
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.3018054329252903e-06,
+      "loss": 0.9351,
+      "step": 3895
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2975217265196804e-06,
+      "loss": 1.0929,
+      "step": 3896
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2932414924690793e-06,
+      "loss": 0.99,
+      "step": 3897
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2889647327030594e-06,
+      "loss": 1.03,
+      "step": 3898
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2846914491496353e-06,
+      "loss": 0.8222,
+      "step": 3899
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.280421643735233e-06,
+      "loss": 1.0324,
+      "step": 3900
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.2761553183847328e-06,
+      "loss": 1.1216,
+      "step": 3901
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.271892475021441e-06,
+      "loss": 1.0988,
+      "step": 3902
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.2676331155670793e-06,
+      "loss": 1.1386,
+      "step": 3903
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.26337724194182e-06,
+      "loss": 1.2333,
+      "step": 3904
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.2591248560642475e-06,
+      "loss": 1.0584,
+      "step": 3905
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.2548759598513815e-06,
+      "loss": 0.9514,
+      "step": 3906
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.25063055521867e-06,
+      "loss": 0.8988,
+      "step": 3907
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.2463886440799833e-06,
+      "loss": 0.9667,
+      "step": 3908
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.242150228347615e-06,
+      "loss": 0.9021,
+      "step": 3909
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.237915309932289e-06,
+      "loss": 0.9376,
+      "step": 3910
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.233683890743149e-06,
+      "loss": 0.8307,
+      "step": 3911
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.2294559726877585e-06,
+      "loss": 0.7539,
+      "step": 3912
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.225231557672111e-06,
+      "loss": 0.879,
+      "step": 3913
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.221010647600611e-06,
+      "loss": 1.1,
+      "step": 3914
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.21679324437609e-06,
+      "loss": 0.9128,
+      "step": 3915
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.2125793498997906e-06,
+      "loss": 0.7407,
+      "step": 3916
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.208368966071387e-06,
+      "loss": 1.098,
+      "step": 3917
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.2041620947889573e-06,
+      "loss": 1.2314,
+      "step": 3918
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1999587379490016e-06,
+      "loss": 0.7995,
+      "step": 3919
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1957588974464407e-06,
+      "loss": 1.1306,
+      "step": 3920
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1915625751746004e-06,
+      "loss": 0.9028,
+      "step": 3921
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.187369773025223e-06,
+      "loss": 0.9586,
+      "step": 3922
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.183180492888476e-06,
+      "loss": 1.0845,
+      "step": 3923
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1789947366529173e-06,
+      "loss": 1.086,
+      "step": 3924
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1748125062055327e-06,
+      "loss": 0.8745,
+      "step": 3925
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.17063380343172e-06,
+      "loss": 1.057,
+      "step": 3926
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.16645863021527e-06,
+      "loss": 0.8027,
+      "step": 3927
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1622869884384012e-06,
+      "loss": 0.912,
+      "step": 3928
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1581188799817287e-06,
+      "loss": 1.1628,
+      "step": 3929
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1539543067242764e-06,
+      "loss": 1.3786,
+      "step": 3930
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.149793270543479e-06,
+      "loss": 1.3273,
+      "step": 3931
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1456357733151724e-06,
+      "loss": 0.9878,
+      "step": 3932
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.141481816913595e-06,
+      "loss": 1.1106,
+      "step": 3933
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.1373314032114024e-06,
+      "loss": 0.9818,
+      "step": 3934
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.1331845340796298e-06,
+      "loss": 1.2583,
+      "step": 3935
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.1290412113877344e-06,
+      "loss": 1.0173,
+      "step": 3936
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.124901437003574e-06,
+      "loss": 0.9579,
+      "step": 3937
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.1207652127933875e-06,
+      "loss": 1.2427,
+      "step": 3938
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.1166325406218358e-06,
+      "loss": 1.0312,
+      "step": 3939
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.112503422351968e-06,
+      "loss": 0.9057,
+      "step": 3940
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.1083778598452278e-06,
+      "loss": 1.04,
+      "step": 3941
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.1042558549614667e-06,
+      "loss": 1.0904,
+      "step": 3942
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.1001374095589224e-06,
+      "loss": 1.099,
+      "step": 3943
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.09602252549423e-06,
+      "loss": 1.0333,
+      "step": 3944
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0919112046224267e-06,
+      "loss": 1.123,
+      "step": 3945
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0878034487969347e-06,
+      "loss": 1.1068,
+      "step": 3946
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0836992598695704e-06,
+      "loss": 1.1731,
+      "step": 3947
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.079598639690541e-06,
+      "loss": 0.7877,
+      "step": 3948
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.075501590108454e-06,
+      "loss": 0.976,
+      "step": 3949
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.071408112970299e-06,
+      "loss": 1.0325,
+      "step": 3950
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.0673182101214517e-06,
+      "loss": 1.0344,
+      "step": 3951
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.0632318834056887e-06,
+      "loss": 0.9293,
+      "step": 3952
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.0591491346651627e-06,
+      "loss": 0.849,
+      "step": 3953
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.0550699657404205e-06,
+      "loss": 1.0658,
+      "step": 3954
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.050994378470389e-06,
+      "loss": 1.1525,
+      "step": 3955
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.0469223746923896e-06,
+      "loss": 0.9601,
+      "step": 3956
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.042853956242119e-06,
+      "loss": 0.9953,
+      "step": 3957
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.0387891249536628e-06,
+      "loss": 1.0773,
+      "step": 3958
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.0347278826594853e-06,
+      "loss": 0.9235,
+      "step": 3959
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.0306702311904393e-06,
+      "loss": 0.9378,
+      "step": 3960
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.026616172375754e-06,
+      "loss": 1.172,
+      "step": 3961
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.0225657080430374e-06,
+      "loss": 1.0048,
+      "step": 3962
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.018518840018283e-06,
+      "loss": 0.9383,
+      "step": 3963
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.01447557012586e-06,
+      "loss": 1.1419,
+      "step": 3964
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.010435900188511e-06,
+      "loss": 1.1664,
+      "step": 3965
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.006399832027367e-06,
+      "loss": 1.0199,
+      "step": 3966
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.0023673674619205e-06,
+      "loss": 1.1925,
+      "step": 3967
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9983385083100504e-06,
+      "loss": 1.0985,
+      "step": 3968
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.994313256388013e-06,
+      "loss": 1.0155,
+      "step": 3969
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.990291613510421e-06,
+      "loss": 1.1999,
+      "step": 3970
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9862735814902823e-06,
+      "loss": 1.2858,
+      "step": 3971
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9822591621389608e-06,
+      "loss": 1.2561,
+      "step": 3972
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9782483572661957e-06,
+      "loss": 0.945,
+      "step": 3973
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9742411686801034e-06,
+      "loss": 0.9465,
+      "step": 3974
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9702375981871634e-06,
+      "loss": 1.0699,
+      "step": 3975
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.966237647592222e-06,
+      "loss": 1.2314,
+      "step": 3976
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9622413186985034e-06,
+      "loss": 0.9857,
+      "step": 3977
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9582486133075884e-06,
+      "loss": 0.8556,
+      "step": 3978
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.954259533219429e-06,
+      "loss": 0.9375,
+      "step": 3979
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9502740802323473e-06,
+      "loss": 0.9943,
+      "step": 3980
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.946292256143022e-06,
+      "loss": 1.2077,
+      "step": 3981
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9423140627465e-06,
+      "loss": 0.7764,
+      "step": 3982
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9383395018361915e-06,
+      "loss": 1.1032,
+      "step": 3983
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9343685752038653e-06,
+      "loss": 0.9154,
+      "step": 3984
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9304012846396625e-06,
+      "loss": 0.9058,
+      "step": 3985
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.926437631932073e-06,
+      "loss": 1.0968,
+      "step": 3986
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9224776188679506e-06,
+      "loss": 1.0179,
+      "step": 3987
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9185212472325123e-06,
+      "loss": 1.159,
+      "step": 3988
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.91456851880933e-06,
+      "loss": 1.0612,
+      "step": 3989
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9106194353803308e-06,
+      "loss": 0.9068,
+      "step": 3990
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.906673998725802e-06,
+      "loss": 0.8081,
+      "step": 3991
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9027322106243895e-06,
+      "loss": 0.837,
+      "step": 3992
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8987940728530897e-06,
+      "loss": 1.0227,
+      "step": 3993
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8948595871872511e-06,
+      "loss": 1.2257,
+      "step": 3994
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8909287554005862e-06,
+      "loss": 1.0487,
+      "step": 3995
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8870015792651497e-06,
+      "loss": 1.0552,
+      "step": 3996
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8830780605513499e-06,
+      "loss": 1.0195,
+      "step": 3997
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8791582010279553e-06,
+      "loss": 1.0517,
+      "step": 3998
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8752420024620732e-06,
+      "loss": 1.0618,
+      "step": 3999
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.8713294666191651e-06,
+      "loss": 1.0568,
+      "step": 4000
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.8674205952630475e-06,
+      "loss": 0.8858,
+      "step": 4001
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.8635153901558712e-06,
+      "loss": 1.0278,
+      "step": 4002
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.8596138530581453e-06,
+      "loss": 0.8921,
+      "step": 4003
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.8557159857287276e-06,
+      "loss": 1.0788,
+      "step": 4004
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.851821789924806e-06,
+      "loss": 0.8827,
+      "step": 4005
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.8479312674019313e-06,
+      "loss": 1.1686,
+      "step": 4006
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.8440444199139862e-06,
+      "loss": 0.8789,
+      "step": 4007
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.8401612492132015e-06,
+      "loss": 0.8986,
+      "step": 4008
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.8362817570501512e-06,
+      "loss": 0.9244,
+      "step": 4009
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.8324059451737496e-06,
+      "loss": 1.1134,
+      "step": 4010
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.8285338153312493e-06,
+      "loss": 1.0139,
+      "step": 4011
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.8246653692682514e-06,
+      "loss": 0.986,
+      "step": 4012
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.8208006087286811e-06,
+      "loss": 0.9723,
+      "step": 4013
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.816939535454819e-06,
+      "loss": 0.8783,
+      "step": 4014
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.8130821511872743e-06,
+      "loss": 1.0018,
+      "step": 4015
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.809228457664991e-06,
+      "loss": 0.9203,
+      "step": 4016
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.8053784566252597e-06,
+      "loss": 0.9278,
+      "step": 4017
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.8015321498036952e-06,
+      "loss": 1.0268,
+      "step": 4018
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7976895389342497e-06,
+      "loss": 0.8742,
+      "step": 4019
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7938506257492172e-06,
+      "loss": 0.9306,
+      "step": 4020
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7900154119792147e-06,
+      "loss": 0.8363,
+      "step": 4021
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7861838993531933e-06,
+      "loss": 0.9363,
+      "step": 4022
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7823560895984427e-06,
+      "loss": 1.1358,
+      "step": 4023
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7785319844405757e-06,
+      "loss": 0.9451,
+      "step": 4024
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.774711585603537e-06,
+      "loss": 1.2071,
+      "step": 4025
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7708948948095993e-06,
+      "loss": 1.0152,
+      "step": 4026
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7670819137793693e-06,
+      "loss": 0.9778,
+      "step": 4027
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.763272644231776e-06,
+      "loss": 1.2162,
+      "step": 4028
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7594670878840737e-06,
+      "loss": 0.8603,
+      "step": 4029
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.755665246451851e-06,
+      "loss": 1.1041,
+      "step": 4030
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7518671216490125e-06,
+      "loss": 1.2342,
+      "step": 4031
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7480727151877896e-06,
+      "loss": 0.8399,
+      "step": 4032
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.7442820287787443e-06,
+      "loss": 0.9785,
+      "step": 4033
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.740495064130754e-06,
+      "loss": 1.1451,
+      "step": 4034
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.7367118229510172e-06,
+      "loss": 1.2279,
+      "step": 4035
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.7329323069450654e-06,
+      "loss": 1.0716,
+      "step": 4036
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.7291565178167335e-06,
+      "loss": 0.937,
+      "step": 4037
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.725384457268191e-06,
+      "loss": 0.9678,
+      "step": 4038
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.7216161269999199e-06,
+      "loss": 1.2498,
+      "step": 4039
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.7178515287107178e-06,
+      "loss": 1.0683,
+      "step": 4040
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.7140906640977095e-06,
+      "loss": 1.0613,
+      "step": 4041
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.7103335348563277e-06,
+      "loss": 0.953,
+      "step": 4042
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.706580142680323e-06,
+      "loss": 1.0195,
+      "step": 4043
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.7028304892617675e-06,
+      "loss": 0.9949,
+      "step": 4044
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6990845762910358e-06,
+      "loss": 0.9751,
+      "step": 4045
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6953424054568268e-06,
+      "loss": 1.0345,
+      "step": 4046
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6916039784461535e-06,
+      "loss": 1.0515,
+      "step": 4047
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6878692969443267e-06,
+      "loss": 0.9756,
+      "step": 4048
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6841383626349873e-06,
+      "loss": 1.0456,
+      "step": 4049
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.680411177200074e-06,
+      "loss": 0.8963,
+      "step": 4050
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.676687742319838e-06,
+      "loss": 0.8414,
+      "step": 4051
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.672968059672847e-06,
+      "loss": 1.0425,
+      "step": 4052
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6692521309359676e-06,
+      "loss": 1.3066,
+      "step": 4053
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6655399577843766e-06,
+      "loss": 1.1625,
+      "step": 4054
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6618315418915643e-06,
+      "loss": 1.1573,
+      "step": 4055
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6581268849293175e-06,
+      "loss": 1.0173,
+      "step": 4056
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6544259885677338e-06,
+      "loss": 1.0929,
+      "step": 4057
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6507288544752164e-06,
+      "loss": 1.0934,
+      "step": 4058
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.64703548431847e-06,
+      "loss": 0.9039,
+      "step": 4059
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6433458797625034e-06,
+      "loss": 1.1519,
+      "step": 4060
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6396600424706232e-06,
+      "loss": 0.8465,
+      "step": 4061
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6359779741044502e-06,
+      "loss": 0.9643,
+      "step": 4062
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6322996763238918e-06,
+      "loss": 0.8227,
+      "step": 4063
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.628625150787163e-06,
+      "loss": 0.9596,
+      "step": 4064
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6249543991507788e-06,
+      "loss": 1.0748,
+      "step": 4065
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.6212874230695508e-06,
+      "loss": 0.949,
+      "step": 4066
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.6176242241965878e-06,
+      "loss": 1.0918,
+      "step": 4067
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.6139648041832979e-06,
+      "loss": 1.1177,
+      "step": 4068
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.610309164679381e-06,
+      "loss": 1.0566,
+      "step": 4069
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.6066573073328418e-06,
+      "loss": 0.923,
+      "step": 4070
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.6030092337899716e-06,
+      "loss": 1.2476,
+      "step": 4071
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5993649456953563e-06,
+      "loss": 1.0537,
+      "step": 4072
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.595724444691883e-06,
+      "loss": 0.8614,
+      "step": 4073
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.592087732420725e-06,
+      "loss": 0.9182,
+      "step": 4074
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5884548105213438e-06,
+      "loss": 1.0997,
+      "step": 4075
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5848256806315055e-06,
+      "loss": 1.0379,
+      "step": 4076
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5812003443872526e-06,
+      "loss": 1.2743,
+      "step": 4077
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5775788034229234e-06,
+      "loss": 0.8599,
+      "step": 4078
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5739610593711518e-06,
+      "loss": 0.9615,
+      "step": 4079
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5703471138628434e-06,
+      "loss": 1.0517,
+      "step": 4080
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.566736968527206e-06,
+      "loss": 0.7798,
+      "step": 4081
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5631306249917344e-06,
+      "loss": 1.0176,
+      "step": 4082
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.559528084882196e-06,
+      "loss": 0.9567,
+      "step": 4083
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.555929349822659e-06,
+      "loss": 0.967,
+      "step": 4084
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5523344214354664e-06,
+      "loss": 1.192,
+      "step": 4085
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5487433013412466e-06,
+      "loss": 0.8704,
+      "step": 4086
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5451559911589165e-06,
+      "loss": 0.9236,
+      "step": 4087
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5415724925056718e-06,
+      "loss": 1.0982,
+      "step": 4088
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5379928069969852e-06,
+      "loss": 1.1551,
+      "step": 4089
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5344169362466199e-06,
+      "loss": 0.9311,
+      "step": 4090
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5308448818666133e-06,
+      "loss": 0.9106,
+      "step": 4091
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5272766454672838e-06,
+      "loss": 0.917,
+      "step": 4092
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5237122286572247e-06,
+      "loss": 1.1521,
+      "step": 4093
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5201516330433175e-06,
+      "loss": 1.0204,
+      "step": 4094
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5165948602307102e-06,
+      "loss": 1.2047,
+      "step": 4095
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5130419118228335e-06,
+      "loss": 0.8483,
+      "step": 4096
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5094927894213895e-06,
+      "loss": 1.0825,
+      "step": 4097
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5059474946263641e-06,
+      "loss": 1.0747,
+      "step": 4098
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.5024060290360075e-06,
+      "loss": 1.1891,
+      "step": 4099
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4988683942468486e-06,
+      "loss": 1.0462,
+      "step": 4100
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.495334591853691e-06,
+      "loss": 1.1347,
+      "step": 4101
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4918046234496087e-06,
+      "loss": 1.1617,
+      "step": 4102
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.488278490625944e-06,
+      "loss": 1.0498,
+      "step": 4103
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4847561949723132e-06,
+      "loss": 0.9538,
+      "step": 4104
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4812377380766073e-06,
+      "loss": 1.1278,
+      "step": 4105
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4777231215249787e-06,
+      "loss": 1.2653,
+      "step": 4106
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.474212346901851e-06,
+      "loss": 1.0916,
+      "step": 4107
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4707054157899213e-06,
+      "loss": 1.0467,
+      "step": 4108
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4672023297701465e-06,
+      "loss": 1.0021,
+      "step": 4109
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4637030904217509e-06,
+      "loss": 0.8879,
+      "step": 4110
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4602076993222337e-06,
+      "loss": 1.105,
+      "step": 4111
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4567161580473444e-06,
+      "loss": 1.1734,
+      "step": 4112
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4532284681711096e-06,
+      "loss": 1.1615,
+      "step": 4113
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.449744631265818e-06,
+      "loss": 1.0101,
+      "step": 4114
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.4462646489020116e-06,
+      "loss": 1.1599,
+      "step": 4115
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.442788522648506e-06,
+      "loss": 0.8579,
+      "step": 4116
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.4393162540723747e-06,
+      "loss": 1.0934,
+      "step": 4117
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.4358478447389469e-06,
+      "loss": 1.108,
+      "step": 4118
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.4323832962118211e-06,
+      "loss": 0.8393,
+      "step": 4119
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.4289226100528496e-06,
+      "loss": 0.9627,
+      "step": 4120
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.4254657878221423e-06,
+      "loss": 0.9956,
+      "step": 4121
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.4220128310780746e-06,
+      "loss": 1.1326,
+      "step": 4122
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.41856374137727e-06,
+      "loss": 0.9871,
+      "step": 4123
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.415118520274612e-06,
+      "loss": 0.7921,
+      "step": 4124
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.4116771693232477e-06,
+      "loss": 0.9496,
+      "step": 4125
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.4082396900745642e-06,
+      "loss": 1.0943,
+      "step": 4126
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.4048060840782185e-06,
+      "loss": 0.81,
+      "step": 4127
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.4013763528821122e-06,
+      "loss": 0.9206,
+      "step": 4128
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.397950498032401e-06,
+      "loss": 1.0281,
+      "step": 4129
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3945285210734993e-06,
+      "loss": 1.0984,
+      "step": 4130
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3911104235480667e-06,
+      "loss": 0.9261,
+      "step": 4131
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3876962069970123e-06,
+      "loss": 0.8844,
+      "step": 4132
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3842858729595067e-06,
+      "loss": 1.0817,
+      "step": 4133
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3808794229729582e-06,
+      "loss": 0.9803,
+      "step": 4134
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3774768585730303e-06,
+      "loss": 0.7683,
+      "step": 4135
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.374078181293631e-06,
+      "loss": 1.0507,
+      "step": 4136
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3706833926669228e-06,
+      "loss": 0.9493,
+      "step": 4137
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3672924942233068e-06,
+      "loss": 0.9114,
+      "step": 4138
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3639054874914348e-06,
+      "loss": 1.1142,
+      "step": 4139
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3605223739982066e-06,
+      "loss": 0.9758,
+      "step": 4140
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.357143155268762e-06,
+      "loss": 0.9885,
+      "step": 4141
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3537678328264836e-06,
+      "loss": 0.9532,
+      "step": 4142
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3503964081930076e-06,
+      "loss": 1.2467,
+      "step": 4143
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3470288828882016e-06,
+      "loss": 1.0713,
+      "step": 4144
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3436652584301802e-06,
+      "loss": 1.0231,
+      "step": 4145
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3403055363353046e-06,
+      "loss": 0.9149,
+      "step": 4146
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3369497181181613e-06,
+      "loss": 1.0509,
+      "step": 4147
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.3335978052915943e-06,
+      "loss": 1.1512,
+      "step": 4148
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.330249799366682e-06,
+      "loss": 1.1336,
+      "step": 4149
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.326905701852732e-06,
+      "loss": 0.9396,
+      "step": 4150
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.3235655142573023e-06,
+      "loss": 1.0993,
+      "step": 4151
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.3202292380861814e-06,
+      "loss": 0.9035,
+      "step": 4152
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.3168968748433953e-06,
+      "loss": 0.8885,
+      "step": 4153
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.3135684260312109e-06,
+      "loss": 1.0461,
+      "step": 4154
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.3102438931501237e-06,
+      "loss": 0.9736,
+      "step": 4155
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.3069232776988638e-06,
+      "loss": 0.9045,
+      "step": 4156
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.303606581174407e-06,
+      "loss": 0.9461,
+      "step": 4157
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.3002938050719437e-06,
+      "loss": 0.9694,
+      "step": 4158
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.296984950884913e-06,
+      "loss": 1.0714,
+      "step": 4159
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.293680020104977e-06,
+      "loss": 1.1351,
+      "step": 4160
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2903790142220318e-06,
+      "loss": 1.1113,
+      "step": 4161
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2870819347242069e-06,
+      "loss": 0.9692,
+      "step": 4162
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2837887830978579e-06,
+      "loss": 0.8836,
+      "step": 4163
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2804995608275683e-06,
+      "loss": 1.1595,
+      "step": 4164
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2772142693961565e-06,
+      "loss": 1.0768,
+      "step": 4165
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2739329102846632e-06,
+      "loss": 1.1792,
+      "step": 4166
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2706554849723551e-06,
+      "loss": 0.8943,
+      "step": 4167
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2673819949367361e-06,
+      "loss": 1.2975,
+      "step": 4168
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2641124416535222e-06,
+      "loss": 0.9505,
+      "step": 4169
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2608468265966645e-06,
+      "loss": 1.0652,
+      "step": 4170
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2575851512383318e-06,
+      "loss": 1.1484,
+      "step": 4171
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2543274170489238e-06,
+      "loss": 1.0222,
+      "step": 4172
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.251073625497059e-06,
+      "loss": 1.0083,
+      "step": 4173
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2478237780495773e-06,
+      "loss": 1.1623,
+      "step": 4174
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2445778761715465e-06,
+      "loss": 1.0461,
+      "step": 4175
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2413359213262498e-06,
+      "loss": 1.014,
+      "step": 4176
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2380979149751904e-06,
+      "loss": 1.0586,
+      "step": 4177
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2348638585781004e-06,
+      "loss": 1.1286,
+      "step": 4178
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2316337535929224e-06,
+      "loss": 1.1231,
+      "step": 4179
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2284076014758196e-06,
+      "loss": 0.9817,
+      "step": 4180
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.225185403681174e-06,
+      "loss": 1.0776,
+      "step": 4181
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.2219671616615824e-06,
+      "loss": 0.9881,
+      "step": 4182
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.2187528768678668e-06,
+      "loss": 1.0596,
+      "step": 4183
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.2155425507490558e-06,
+      "loss": 1.1616,
+      "step": 4184
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.212336184752394e-06,
+      "loss": 1.1506,
+      "step": 4185
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.2091337803233505e-06,
+      "loss": 0.9803,
+      "step": 4186
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.2059353389055962e-06,
+      "loss": 0.912,
+      "step": 4187
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.2027408619410208e-06,
+      "loss": 0.9383,
+      "step": 4188
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1995503508697326e-06,
+      "loss": 0.9702,
+      "step": 4189
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1963638071300388e-06,
+      "loss": 0.9714,
+      "step": 4190
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.193181232158467e-06,
+      "loss": 1.0661,
+      "step": 4191
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1900026273897602e-06,
+      "loss": 0.9364,
+      "step": 4192
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1868279942568584e-06,
+      "loss": 1.157,
+      "step": 4193
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1836573341909219e-06,
+      "loss": 0.7841,
+      "step": 4194
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1804906486213153e-06,
+      "loss": 0.9342,
+      "step": 4195
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.177327938975611e-06,
+      "loss": 1.1858,
+      "step": 4196
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.174169206679593e-06,
+      "loss": 1.3148,
+      "step": 4197
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1710144531572499e-06,
+      "loss": 0.9406,
+      "step": 4198
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.167863679830773e-06,
+      "loss": 0.9766,
+      "step": 4199
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1647168881205672e-06,
+      "loss": 0.9354,
+      "step": 4200
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1615740794452358e-06,
+      "loss": 0.9121,
+      "step": 4201
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.158435255221586e-06,
+      "loss": 0.8662,
+      "step": 4202
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1553004168646364e-06,
+      "loss": 1.0102,
+      "step": 4203
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.152169565787602e-06,
+      "loss": 1.0267,
+      "step": 4204
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1490427034019014e-06,
+      "loss": 1.0659,
+      "step": 4205
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1459198311171537e-06,
+      "loss": 1.0447,
+      "step": 4206
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1428009503411863e-06,
+      "loss": 0.8974,
+      "step": 4207
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1396860624800188e-06,
+      "loss": 0.9086,
+      "step": 4208
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1365751689378757e-06,
+      "loss": 0.9416,
+      "step": 4209
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1334682711171752e-06,
+      "loss": 1.0481,
+      "step": 4210
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1303653704185435e-06,
+      "loss": 0.8996,
+      "step": 4211
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1272664682407985e-06,
+      "loss": 0.983,
+      "step": 4212
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1241715659809538e-06,
+      "loss": 0.9028,
+      "step": 4213
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.121080665034223e-06,
+      "loss": 1.0783,
+      "step": 4214
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.1179937667940188e-06,
+      "loss": 0.9027,
+      "step": 4215
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.1149108726519443e-06,
+      "loss": 0.9127,
+      "step": 4216
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.1118319839977976e-06,
+      "loss": 0.9228,
+      "step": 4217
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.1087571022195754e-06,
+      "loss": 0.9545,
+      "step": 4218
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.1056862287034641e-06,
+      "loss": 1.1767,
+      "step": 4219
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.1026193648338436e-06,
+      "loss": 1.0036,
+      "step": 4220
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0995565119932905e-06,
+      "loss": 1.0356,
+      "step": 4221
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0964976715625664e-06,
+      "loss": 1.1191,
+      "step": 4222
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.093442844920628e-06,
+      "loss": 1.0212,
+      "step": 4223
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.090392033444625e-06,
+      "loss": 0.9485,
+      "step": 4224
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0873452385098892e-06,
+      "loss": 1.0173,
+      "step": 4225
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0843024614899479e-06,
+      "loss": 0.9108,
+      "step": 4226
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.081263703756522e-06,
+      "loss": 0.8236,
+      "step": 4227
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0782289666795042e-06,
+      "loss": 1.0048,
+      "step": 4228
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.075198251626991e-06,
+      "loss": 0.8577,
+      "step": 4229
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0721715599652581e-06,
+      "loss": 1.2365,
+      "step": 4230
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0691488930587668e-06,
+      "loss": 1.0349,
+      "step": 4231
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.06613025227017e-06,
+      "loss": 0.9731,
+      "step": 4232
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0631156389602982e-06,
+      "loss": 1.0897,
+      "step": 4233
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0601050544881675e-06,
+      "loss": 1.3526,
+      "step": 4234
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0570985002109856e-06,
+      "loss": 1.3179,
+      "step": 4235
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.054095977484134e-06,
+      "loss": 1.126,
+      "step": 4236
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0510974876611813e-06,
+      "loss": 0.9867,
+      "step": 4237
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0481030320938756e-06,
+      "loss": 1.0372,
+      "step": 4238
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0451126121321475e-06,
+      "loss": 0.8871,
+      "step": 4239
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0421262291241108e-06,
+      "loss": 0.988,
+      "step": 4240
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.039143884416056e-06,
+      "loss": 1.0594,
+      "step": 4241
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0361655793524528e-06,
+      "loss": 1.0998,
+      "step": 4242
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0331913152759532e-06,
+      "loss": 1.0877,
+      "step": 4243
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.030221093527386e-06,
+      "loss": 0.9937,
+      "step": 4244
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0272549154457533e-06,
+      "loss": 0.8633,
+      "step": 4245
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.0242927823682435e-06,
+      "loss": 0.8636,
+      "step": 4246
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.0213346956302129e-06,
+      "loss": 1.2648,
+      "step": 4247
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.0183806565651978e-06,
+      "loss": 0.9513,
+      "step": 4248
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.0154306665049063e-06,
+      "loss": 1.1259,
+      "step": 4249
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.012484726779228e-06,
+      "loss": 1.0819,
+      "step": 4250
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.00954283871622e-06,
+      "loss": 0.8376,
+      "step": 4251
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.0066050036421137e-06,
+      "loss": 0.994,
+      "step": 4252
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.0036712228813183e-06,
+      "loss": 0.802,
+      "step": 4253
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.00074149775641e-06,
+      "loss": 0.8344,
+      "step": 4254
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.978158295881358e-07,
+      "loss": 0.7788,
+      "step": 4255
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.948942196954237e-07,
+      "loss": 0.9758,
+      "step": 4256
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.919766693953558e-07,
+      "loss": 0.9856,
+      "step": 4257
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.89063180003199e-07,
+      "loss": 0.9796,
+      "step": 4258
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.861537528323862e-07,
+      "loss": 0.9581,
+      "step": 4259
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.832483891945076e-07,
+      "loss": 1.0572,
+      "step": 4260
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.803470903993395e-07,
+      "loss": 1.0581,
+      "step": 4261
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.774498577548119e-07,
+      "loss": 0.8567,
+      "step": 4262
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.745566925670259e-07,
+      "loss": 0.9238,
+      "step": 4263
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.716675961402533e-07,
+      "loss": 0.9796,
+      "step": 4264
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.687825697769271e-07,
+      "loss": 1.1169,
+      "step": 4265
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.659016147776424e-07,
+      "loss": 1.1082,
+      "step": 4266
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.63024732441169e-07,
+      "loss": 0.941,
+      "step": 4267
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.601519240644275e-07,
+      "loss": 0.9898,
+      "step": 4268
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.57283190942513e-07,
+      "loss": 0.9361,
+      "step": 4269
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.54418534368682e-07,
+      "loss": 1.059,
+      "step": 4270
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.515579556343446e-07,
+      "loss": 1.0369,
+      "step": 4271
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.48701456029083e-07,
+      "loss": 1.0669,
+      "step": 4272
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.458490368406348e-07,
+      "loss": 1.2346,
+      "step": 4273
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.430006993548968e-07,
+      "loss": 0.9796,
+      "step": 4274
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.40156444855933e-07,
+      "loss": 1.0312,
+      "step": 4275
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.373162746259601e-07,
+      "loss": 0.9379,
+      "step": 4276
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.344801899453537e-07,
+      "loss": 0.9136,
+      "step": 4277
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.316481920926512e-07,
+      "loss": 1.0616,
+      "step": 4278
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.28820282344547e-07,
+      "loss": 0.9332,
+      "step": 4279
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.259964619758877e-07,
+      "loss": 0.8291,
+      "step": 4280
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.231767322596841e-07,
+      "loss": 1.0755,
+      "step": 4281
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.203610944670971e-07,
+      "loss": 0.9759,
+      "step": 4282
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.175495498674447e-07,
+      "loss": 0.8608,
+      "step": 4283
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.147420997281975e-07,
+      "loss": 0.9088,
+      "step": 4284
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.119387453149864e-07,
+      "loss": 1.0161,
+      "step": 4285
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.091394878915893e-07,
+      "loss": 0.8961,
+      "step": 4286
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.063443287199391e-07,
+      "loss": 1.0713,
+      "step": 4287
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.035532690601245e-07,
+      "loss": 0.9638,
+      "step": 4288
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.007663101703823e-07,
+      "loss": 0.9298,
+      "step": 4289
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.979834533070986e-07,
+      "loss": 0.9393,
+      "step": 4290
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.952046997248199e-07,
+      "loss": 0.9538,
+      "step": 4291
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.924300506762296e-07,
+      "loss": 1.0017,
+      "step": 4292
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.896595074121716e-07,
+      "loss": 1.0439,
+      "step": 4293
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.868930711816348e-07,
+      "loss": 1.1786,
+      "step": 4294
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.841307432317525e-07,
+      "loss": 1.0473,
+      "step": 4295
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.813725248078153e-07,
+      "loss": 0.8964,
+      "step": 4296
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.78618417153253e-07,
+      "loss": 1.0903,
+      "step": 4297
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.758684215096446e-07,
+      "loss": 0.7728,
+      "step": 4298
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.731225391167187e-07,
+      "loss": 1.2819,
+      "step": 4299
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.703807712123457e-07,
+      "loss": 1.1811,
+      "step": 4300
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.676431190325396e-07,
+      "loss": 1.1233,
+      "step": 4301
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.649095838114674e-07,
+      "loss": 0.8493,
+      "step": 4302
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.621801667814267e-07,
+      "loss": 1.0724,
+      "step": 4303
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.59454869172871e-07,
+      "loss": 0.9441,
+      "step": 4304
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.567336922143954e-07,
+      "loss": 1.0169,
+      "step": 4305
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.540166371327264e-07,
+      "loss": 1.0734,
+      "step": 4306
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.51303705152744e-07,
+      "loss": 1.0241,
+      "step": 4307
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.485948974974656e-07,
+      "loss": 1.1059,
+      "step": 4308
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.458902153880466e-07,
+      "loss": 1.0135,
+      "step": 4309
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.431896600437883e-07,
+      "loss": 0.8787,
+      "step": 4310
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.404932326821257e-07,
+      "loss": 0.9801,
+      "step": 4311
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.378009345186356e-07,
+      "loss": 1.1998,
+      "step": 4312
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.351127667670355e-07,
+      "loss": 1.1897,
+      "step": 4313
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.324287306391776e-07,
+      "loss": 1.1789,
+      "step": 4314
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.297488273450515e-07,
+      "loss": 1.2339,
+      "step": 4315
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.270730580927844e-07,
+      "loss": 0.9643,
+      "step": 4316
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.244014240886422e-07,
+      "loss": 1.0928,
+      "step": 4317
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.217339265370227e-07,
+      "loss": 1.0742,
+      "step": 4318
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.190705666404597e-07,
+      "loss": 1.023,
+      "step": 4319
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.164113455996258e-07,
+      "loss": 0.9111,
+      "step": 4320
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.137562646133246e-07,
+      "loss": 1.077,
+      "step": 4321
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.111053248784906e-07,
+      "loss": 0.9456,
+      "step": 4322
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.084585275901946e-07,
+      "loss": 1.0739,
+      "step": 4323
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.058158739416421e-07,
+      "loss": 1.1673,
+      "step": 4324
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.031773651241659e-07,
+      "loss": 0.9322,
+      "step": 4325
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.005430023272343e-07,
+      "loss": 0.9767,
+      "step": 4326
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.979127867384406e-07,
+      "loss": 1.2079,
+      "step": 4327
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.952867195435177e-07,
+      "loss": 1.2693,
+      "step": 4328
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.926648019263216e-07,
+      "loss": 1.1387,
+      "step": 4329
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.90047035068836e-07,
+      "loss": 1.0213,
+      "step": 4330
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.874334201511813e-07,
+      "loss": 0.9863,
+      "step": 4331
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.848239583515982e-07,
+      "loss": 1.0106,
+      "step": 4332
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.822186508464591e-07,
+      "loss": 0.9533,
+      "step": 4333
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.796174988102667e-07,
+      "loss": 0.9813,
+      "step": 4334
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.770205034156408e-07,
+      "loss": 1.0876,
+      "step": 4335
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.74427665833335e-07,
+      "loss": 0.8851,
+      "step": 4336
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.718389872322307e-07,
+      "loss": 0.9991,
+      "step": 4337
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.692544687793236e-07,
+      "loss": 0.9453,
+      "step": 4338
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.666741116397458e-07,
+      "loss": 1.2447,
+      "step": 4339
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.640979169767482e-07,
+      "loss": 1.1985,
+      "step": 4340
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.615258859517005e-07,
+      "loss": 0.8285,
+      "step": 4341
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.589580197241065e-07,
+      "loss": 1.1453,
+      "step": 4342
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.563943194515832e-07,
+      "loss": 1.0195,
+      "step": 4343
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.538347862898709e-07,
+      "loss": 0.9693,
+      "step": 4344
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.512794213928365e-07,
+      "loss": 1.1289,
+      "step": 4345
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.487282259124629e-07,
+      "loss": 1.0414,
+      "step": 4346
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.461812009988534e-07,
+      "loss": 1.0549,
+      "step": 4347
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.436383478002352e-07,
+      "loss": 1.0878,
+      "step": 4348
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.410996674629511e-07,
+      "loss": 1.001,
+      "step": 4349
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.385651611314637e-07,
+      "loss": 1.0638,
+      "step": 4350
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.360348299483533e-07,
+      "loss": 1.0009,
+      "step": 4351
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.335086750543174e-07,
+      "loss": 1.0208,
+      "step": 4352
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.309866975881763e-07,
+      "loss": 1.04,
+      "step": 4353
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.284688986868616e-07,
+      "loss": 0.9653,
+      "step": 4354
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.259552794854186e-07,
+      "loss": 0.9217,
+      "step": 4355
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.234458411170176e-07,
+      "loss": 0.9829,
+      "step": 4356
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.209405847129358e-07,
+      "loss": 1.0334,
+      "step": 4357
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.184395114025689e-07,
+      "loss": 0.8209,
+      "step": 4358
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.159426223134247e-07,
+      "loss": 1.1949,
+      "step": 4359
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.134499185711285e-07,
+      "loss": 0.9946,
+      "step": 4360
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 7.109614012994148e-07,
+      "loss": 1.0432,
+      "step": 4361
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 7.084770716201306e-07,
+      "loss": 1.0057,
+      "step": 4362
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 7.05996930653241e-07,
+      "loss": 1.0824,
+      "step": 4363
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 7.035209795168163e-07,
+      "loss": 1.2285,
+      "step": 4364
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 7.010492193270379e-07,
+      "loss": 1.0973,
+      "step": 4365
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.985816511982047e-07,
+      "loss": 0.9984,
+      "step": 4366
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.961182762427188e-07,
+      "loss": 1.0378,
+      "step": 4367
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.936590955710943e-07,
+      "loss": 0.9038,
+      "step": 4368
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.912041102919576e-07,
+      "loss": 1.0643,
+      "step": 4369
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.887533215120357e-07,
+      "loss": 0.9888,
+      "step": 4370
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.863067303361726e-07,
+      "loss": 0.9719,
+      "step": 4371
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.838643378673182e-07,
+      "loss": 0.9857,
+      "step": 4372
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.814261452065218e-07,
+      "loss": 1.0587,
+      "step": 4373
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.789921534529509e-07,
+      "loss": 1.0969,
+      "step": 4374
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.765623637038698e-07,
+      "loss": 0.9612,
+      "step": 4375
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.741367770546537e-07,
+      "loss": 1.0321,
+      "step": 4376
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.717153945987842e-07,
+      "loss": 1.1352,
+      "step": 4377
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.692982174278384e-07,
+      "loss": 1.1333,
+      "step": 4378
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.66885246631509e-07,
+      "loss": 1.2522,
+      "step": 4379
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.644764832975892e-07,
+      "loss": 0.874,
+      "step": 4380
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.620719285119693e-07,
+      "loss": 0.981,
+      "step": 4381
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.596715833586498e-07,
+      "loss": 1.0125,
+      "step": 4382
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.572754489197297e-07,
+      "loss": 0.9906,
+      "step": 4383
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.548835262754083e-07,
+      "loss": 1.0964,
+      "step": 4384
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.524958165039929e-07,
+      "loss": 1.0112,
+      "step": 4385
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.50112320681886e-07,
+      "loss": 0.9173,
+      "step": 4386
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.47733039883588e-07,
+      "loss": 0.9767,
+      "step": 4387
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.453579751817074e-07,
+      "loss": 1.0209,
+      "step": 4388
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.429871276469457e-07,
+      "loss": 0.8312,
+      "step": 4389
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.406204983481024e-07,
+      "loss": 1.0662,
+      "step": 4390
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.382580883520817e-07,
+      "loss": 0.892,
+      "step": 4391
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.358998987238807e-07,
+      "loss": 1.1053,
+      "step": 4392
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.335459305265946e-07,
+      "loss": 0.9285,
+      "step": 4393
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.311961848214143e-07,
+      "loss": 1.0304,
+      "step": 4394
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.288506626676316e-07,
+      "loss": 1.0228,
+      "step": 4395
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.265093651226317e-07,
+      "loss": 1.0812,
+      "step": 4396
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.241722932418926e-07,
+      "loss": 1.1559,
+      "step": 4397
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.218394480789925e-07,
+      "loss": 0.8466,
+      "step": 4398
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.195108306856012e-07,
+      "loss": 0.9696,
+      "step": 4399
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.171864421114792e-07,
+      "loss": 1.0856,
+      "step": 4400
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.148662834044905e-07,
+      "loss": 1.147,
+      "step": 4401
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.125503556105827e-07,
+      "loss": 0.7814,
+      "step": 4402
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.102386597737975e-07,
+      "loss": 1.0252,
+      "step": 4403
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.079311969362767e-07,
+      "loss": 1.0215,
+      "step": 4404
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.056279681382416e-07,
+      "loss": 1.0661,
+      "step": 4405
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.033289744180137e-07,
+      "loss": 0.9076,
+      "step": 4406
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.010342168120031e-07,
+      "loss": 1.0759,
+      "step": 4407
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.987436963547077e-07,
+      "loss": 0.9688,
+      "step": 4408
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.964574140787205e-07,
+      "loss": 0.8753,
+      "step": 4409
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.94175371014718e-07,
+      "loss": 1.1347,
+      "step": 4410
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.918975681914685e-07,
+      "loss": 0.9624,
+      "step": 4411
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.896240066358306e-07,
+      "loss": 1.0822,
+      "step": 4412
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.873546873727454e-07,
+      "loss": 0.8765,
+      "step": 4413
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.850896114252469e-07,
+      "loss": 1.1466,
+      "step": 4414
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.828287798144561e-07,
+      "loss": 1.0689,
+      "step": 4415
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.805721935595742e-07,
+      "loss": 0.9209,
+      "step": 4416
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.783198536778978e-07,
+      "loss": 1.1704,
+      "step": 4417
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.76071761184801e-07,
+      "loss": 1.1127,
+      "step": 4418
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.738279170937478e-07,
+      "loss": 1.1505,
+      "step": 4419
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.715883224162866e-07,
+      "loss": 0.9241,
+      "step": 4420
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.693529781620477e-07,
+      "loss": 0.9751,
+      "step": 4421
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.671218853387472e-07,
+      "loss": 1.0444,
+      "step": 4422
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.648950449521873e-07,
+      "loss": 1.0314,
+      "step": 4423
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.626724580062459e-07,
+      "loss": 1.309,
+      "step": 4424
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.604541255028895e-07,
+      "loss": 0.7286,
+      "step": 4425
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.582400484421647e-07,
+      "loss": 1.057,
+      "step": 4426
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.560302278222007e-07,
+      "loss": 1.0036,
+      "step": 4427
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.538246646392065e-07,
+      "loss": 1.2555,
+      "step": 4428
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.516233598874698e-07,
+      "loss": 0.9619,
+      "step": 4429
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.494263145593648e-07,
+      "loss": 0.9862,
+      "step": 4430
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.47233529645339e-07,
+      "loss": 1.0172,
+      "step": 4431
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.45045006133923e-07,
+      "loss": 0.9911,
+      "step": 4432
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.428607450117263e-07,
+      "loss": 0.9385,
+      "step": 4433
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.406807472634346e-07,
+      "loss": 1.0361,
+      "step": 4434
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.385050138718129e-07,
+      "loss": 1.1,
+      "step": 4435
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.363335458177044e-07,
+      "loss": 0.966,
+      "step": 4436
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.341663440800271e-07,
+      "loss": 1.0269,
+      "step": 4437
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.320034096357796e-07,
+      "loss": 0.9706,
+      "step": 4438
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.29844743460034e-07,
+      "loss": 0.9476,
+      "step": 4439
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.276903465259375e-07,
+      "loss": 1.0512,
+      "step": 4440
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.255402198047165e-07,
+      "loss": 1.0726,
+      "step": 4441
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.233943642656692e-07,
+      "loss": 0.9555,
+      "step": 4442
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.212527808761658e-07,
+      "loss": 1.0312,
+      "step": 4443
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.191154706016588e-07,
+      "loss": 1.0558,
+      "step": 4444
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.16982434405665e-07,
+      "loss": 1.0023,
+      "step": 4445
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.1485367324978e-07,
+      "loss": 0.963,
+      "step": 4446
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.127291880936736e-07,
+      "loss": 0.8274,
+      "step": 4447
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.106089798950809e-07,
+      "loss": 1.0404,
+      "step": 4448
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.084930496098139e-07,
+      "loss": 0.8595,
+      "step": 4449
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.063813981917587e-07,
+      "loss": 0.8602,
+      "step": 4450
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.042740265928647e-07,
+      "loss": 0.9009,
+      "step": 4451
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.021709357631599e-07,
+      "loss": 0.8492,
+      "step": 4452
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.000721266507369e-07,
+      "loss": 1.0554,
+      "step": 4453
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.9797760020176e-07,
+      "loss": 0.9821,
+      "step": 4454
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.958873573604639e-07,
+      "loss": 1.3202,
+      "step": 4455
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.938013990691515e-07,
+      "loss": 1.06,
+      "step": 4456
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.917197262681905e-07,
+      "loss": 0.8395,
+      "step": 4457
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.896423398960248e-07,
+      "loss": 1.1148,
+      "step": 4458
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.875692408891597e-07,
+      "loss": 1.0956,
+      "step": 4459
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.855004301821675e-07,
+      "loss": 0.8421,
+      "step": 4460
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.834359087076901e-07,
+      "loss": 0.9527,
+      "step": 4461
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.813756773964328e-07,
+      "loss": 0.9639,
+      "step": 4462
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.793197371771718e-07,
+      "loss": 0.8856,
+      "step": 4463
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.772680889767444e-07,
+      "loss": 1.3103,
+      "step": 4464
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.7522073372005207e-07,
+      "loss": 0.9628,
+      "step": 4465
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.731776723300674e-07,
+      "loss": 0.8658,
+      "step": 4466
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.711389057278215e-07,
+      "loss": 0.9373,
+      "step": 4467
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.691044348324092e-07,
+      "loss": 1.2012,
+      "step": 4468
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6707426056099236e-07,
+      "loss": 0.8875,
+      "step": 4469
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6504838382879337e-07,
+      "loss": 1.1604,
+      "step": 4470
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.630268055490983e-07,
+      "loss": 0.949,
+      "step": 4471
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.610095266332537e-07,
+      "loss": 1.0478,
+      "step": 4472
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5899654799067215e-07,
+      "loss": 1.0819,
+      "step": 4473
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.569878705288222e-07,
+      "loss": 0.9194,
+      "step": 4474
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5498349515323614e-07,
+      "loss": 1.2883,
+      "step": 4475
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.529834227675078e-07,
+      "loss": 0.768,
+      "step": 4476
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.5098765427329047e-07,
+      "loss": 0.9538,
+      "step": 4477
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.489961905702944e-07,
+      "loss": 1.0574,
+      "step": 4478
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.470090325562948e-07,
+      "loss": 0.9991,
+      "step": 4479
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.450261811271184e-07,
+      "loss": 1.1154,
+      "step": 4480
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.4304763717665676e-07,
+      "loss": 1.0626,
+      "step": 4481
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.4107340159685983e-07,
+      "loss": 1.1155,
+      "step": 4482
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.39103475277729e-07,
+      "loss": 0.9416,
+      "step": 4483
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.371378591073294e-07,
+      "loss": 0.9121,
+      "step": 4484
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3517655397178004e-07,
+      "loss": 0.9863,
+      "step": 4485
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.332195607552558e-07,
+      "loss": 0.9722,
+      "step": 4486
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3126688033999107e-07,
+      "loss": 0.9721,
+      "step": 4487
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.293185136062739e-07,
+      "loss": 0.861,
+      "step": 4488
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2737446143244487e-07,
+      "loss": 1.0036,
+      "step": 4489
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.254347246949075e-07,
+      "loss": 1.0643,
+      "step": 4490
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.234993042681079e-07,
+      "loss": 1.1338,
+      "step": 4491
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.215682010245581e-07,
+      "loss": 1.0788,
+      "step": 4492
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.1964141583481944e-07,
+      "loss": 0.9365,
+      "step": 4493
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.177189495675016e-07,
+      "loss": 0.9412,
+      "step": 4494
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.1580080308927574e-07,
+      "loss": 1.0204,
+      "step": 4495
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.138869772648613e-07,
+      "loss": 1.0285,
+      "step": 4496
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.1197747295702695e-07,
+      "loss": 0.8613,
+      "step": 4497
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.100722910266008e-07,
+      "loss": 1.0668,
+      "step": 4498
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.081714323324559e-07,
+      "loss": 1.3283,
+      "step": 4499
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.0627489773151896e-07,
+      "loss": 1.0944,
+      "step": 4500
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.043826880787671e-07,
+      "loss": 1.1757,
+      "step": 4501
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.0249480422722697e-07,
+      "loss": 0.8819,
+      "step": 4502
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.006112470279766e-07,
+      "loss": 1.024,
+      "step": 4503
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9873201733014124e-07,
+      "loss": 1.0637,
+      "step": 4504
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.968571159808987e-07,
+      "loss": 1.08,
+      "step": 4505
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9498654382547185e-07,
+      "loss": 0.9339,
+      "step": 4506
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.931203017071328e-07,
+      "loss": 1.133,
+      "step": 4507
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9125839046720406e-07,
+      "loss": 1.048,
+      "step": 4508
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.894008109450542e-07,
+      "loss": 1.2559,
+      "step": 4509
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.875475639780968e-07,
+      "loss": 0.909,
+      "step": 4510
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.856986504017968e-07,
+      "loss": 0.9626,
+      "step": 4511
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.838540710496608e-07,
+      "loss": 1.0182,
+      "step": 4512
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.820138267532447e-07,
+      "loss": 1.1567,
+      "step": 4513
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.801779183421517e-07,
+      "loss": 0.8396,
+      "step": 4514
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.783463466440229e-07,
+      "loss": 1.0298,
+      "step": 4515
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.7651911248455133e-07,
+      "loss": 0.8915,
+      "step": 4516
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.746962166874768e-07,
+      "loss": 1.0646,
+      "step": 4517
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.7287766007457206e-07,
+      "loss": 0.8964,
+      "step": 4518
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.710634434656657e-07,
+      "loss": 1.06,
+      "step": 4519
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6925356767862466e-07,
+      "loss": 0.9359,
+      "step": 4520
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.674480335293562e-07,
+      "loss": 1.0001,
+      "step": 4521
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6564684183181597e-07,
+      "loss": 0.8786,
+      "step": 4522
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6384999339799886e-07,
+      "loss": 1.07,
+      "step": 4523
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6205748903794136e-07,
+      "loss": 1.2143,
+      "step": 4524
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.602693295597248e-07,
+      "loss": 1.1251,
+      "step": 4525
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.5848551576946666e-07,
+      "loss": 1.045,
+      "step": 4526
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.567060484713303e-07,
+      "loss": 0.8848,
+      "step": 4527
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.549309284675162e-07,
+      "loss": 0.9682,
+      "step": 4528
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.531601565582665e-07,
+      "loss": 0.8487,
+      "step": 4529
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.5139373354186377e-07,
+      "loss": 0.9412,
+      "step": 4530
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.496316602146299e-07,
+      "loss": 1.1511,
+      "step": 4531
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4787393737092387e-07,
+      "loss": 1.2353,
+      "step": 4532
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4612056580314633e-07,
+      "loss": 0.9804,
+      "step": 4533
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4437154630173384e-07,
+      "loss": 0.9212,
+      "step": 4534
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.426268796551613e-07,
+      "loss": 1.051,
+      "step": 4535
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.40886566649945e-07,
+      "loss": 0.9319,
+      "step": 4536
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3915060807063307e-07,
+      "loss": 1.0298,
+      "step": 4537
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.374190046998138e-07,
+      "loss": 0.9646,
+      "step": 4538
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3569175731811067e-07,
+      "loss": 1.0441,
+      "step": 4539
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.339688667041863e-07,
+      "loss": 1.1636,
+      "step": 4540
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.32250333634736e-07,
+      "loss": 0.7635,
+      "step": 4541
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.305361588844913e-07,
+      "loss": 1.2056,
+      "step": 4542
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.2882634322622067e-07,
+      "loss": 1.2006,
+      "step": 4543
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.271208874307252e-07,
+      "loss": 0.9432,
+      "step": 4544
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.2541979226684116e-07,
+      "loss": 0.8417,
+      "step": 4545
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.237230585014417e-07,
+      "loss": 1.2515,
+      "step": 4546
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.220306868994305e-07,
+      "loss": 0.9699,
+      "step": 4547
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.2034267822374396e-07,
+      "loss": 1.1265,
+      "step": 4548
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1865903323535674e-07,
+      "loss": 1.0245,
+      "step": 4549
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1697975269326833e-07,
+      "loss": 0.9708,
+      "step": 4550
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1530483735452087e-07,
+      "loss": 1.0489,
+      "step": 4551
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.136342879741794e-07,
+      "loss": 0.9873,
+      "step": 4552
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.119681053053447e-07,
+      "loss": 0.8899,
+      "step": 4553
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1030629009915047e-07,
+      "loss": 1.0445,
+      "step": 4554
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0864884310475964e-07,
+      "loss": 0.9048,
+      "step": 4555
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0699576506936334e-07,
+      "loss": 1.2034,
+      "step": 4556
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0534705673819e-07,
+      "loss": 0.7503,
+      "step": 4557
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.0370271885449056e-07,
+      "loss": 1.1592,
+      "step": 4558
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.0206275215954985e-07,
+      "loss": 0.9688,
+      "step": 4559
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.0042715739268316e-07,
+      "loss": 1.0237,
+      "step": 4560
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.9879593529123065e-07,
+      "loss": 0.9752,
+      "step": 4561
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.971690865905663e-07,
+      "loss": 0.9539,
+      "step": 4562
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.95546612024088e-07,
+      "loss": 0.8589,
+      "step": 4563
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.9392851232322384e-07,
+      "loss": 0.9791,
+      "step": 4564
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.923147882174304e-07,
+      "loss": 0.9972,
+      "step": 4565
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.907054404341914e-07,
+      "loss": 0.8491,
+      "step": 4566
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8910046969901407e-07,
+      "loss": 0.8257,
+      "step": 4567
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8749987673543866e-07,
+      "loss": 1.1382,
+      "step": 4568
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8590366226502685e-07,
+      "loss": 0.955,
+      "step": 4569
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.843118270073686e-07,
+      "loss": 1.0204,
+      "step": 4570
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.827243716800809e-07,
+      "loss": 0.9384,
+      "step": 4571
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8114129699880254e-07,
+      "loss": 0.9424,
+      "step": 4572
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.795626036772003e-07,
+      "loss": 1.0828,
+      "step": 4573
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.77988292426965e-07,
+      "loss": 1.0555,
+      "step": 4574
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.764183639578122e-07,
+      "loss": 0.9661,
+      "step": 4575
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.7485281897748127e-07,
+      "loss": 1.1099,
+      "step": 4576
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.7329165819173556e-07,
+      "loss": 0.992,
+      "step": 4577
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.7173488230436306e-07,
+      "loss": 0.9382,
+      "step": 4578
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.701824920171725e-07,
+      "loss": 1.0598,
+      "step": 4579
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.686344880299996e-07,
+      "loss": 1.2139,
+      "step": 4580
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.670908710406983e-07,
+      "loss": 0.8335,
+      "step": 4581
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.655516417451454e-07,
+      "loss": 1.1118,
+      "step": 4582
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6401680083724457e-07,
+      "loss": 0.7807,
+      "step": 4583
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6248634900891466e-07,
+      "loss": 0.8702,
+      "step": 4584
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6096028695010045e-07,
+      "loss": 1.0426,
+      "step": 4585
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.594386153487649e-07,
+      "loss": 1.2008,
+      "step": 4586
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.579213348908949e-07,
+      "loss": 1.302,
+      "step": 4587
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.564084462604921e-07,
+      "loss": 1.2575,
+      "step": 4588
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5489995013958655e-07,
+      "loss": 1.0749,
+      "step": 4589
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.53395847208221e-07,
+      "loss": 1.1794,
+      "step": 4590
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.5189613814445956e-07,
+      "loss": 0.8942,
+      "step": 4591
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.504008236243893e-07,
+      "loss": 0.8867,
+      "step": 4592
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.4890990432210883e-07,
+      "loss": 1.0329,
+      "step": 4593
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.474233809097426e-07,
+      "loss": 1.0828,
+      "step": 4594
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.459412540574313e-07,
+      "loss": 0.9724,
+      "step": 4595
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.4446352443332933e-07,
+      "loss": 1.1579,
+      "step": 4596
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.429901927036138e-07,
+      "loss": 1.1351,
+      "step": 4597
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.4152125953247786e-07,
+      "loss": 1.1361,
+      "step": 4598
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.4005672558213067e-07,
+      "loss": 1.0022,
+      "step": 4599
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.385965915128008e-07,
+      "loss": 1.0727,
+      "step": 4600
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3714085798272945e-07,
+      "loss": 1.2088,
+      "step": 4601
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3568952564817615e-07,
+      "loss": 1.0647,
+      "step": 4602
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.342425951634175e-07,
+      "loss": 1.0822,
+      "step": 4603
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3280006718074177e-07,
+      "loss": 1.4306,
+      "step": 4604
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3136194235045872e-07,
+      "loss": 0.9449,
+      "step": 4605
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2992822132088755e-07,
+      "loss": 0.6975,
+      "step": 4606
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2849890473836234e-07,
+      "loss": 1.1825,
+      "step": 4607
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.2707399324723765e-07,
+      "loss": 0.8775,
+      "step": 4608
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.2565348748987527e-07,
+      "loss": 0.9131,
+      "step": 4609
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.2423738810665286e-07,
+      "loss": 0.8552,
+      "step": 4610
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.2282569573596423e-07,
+      "loss": 1.1449,
+      "step": 4611
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.214184110142148e-07,
+      "loss": 1.2889,
+      "step": 4612
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.2001553457582036e-07,
+      "loss": 1.1287,
+      "step": 4613
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1861706705321395e-07,
+      "loss": 1.0418,
+      "step": 4614
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.172230090768379e-07,
+      "loss": 1.0009,
+      "step": 4615
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1583336127514842e-07,
+      "loss": 1.0586,
+      "step": 4616
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1444812427461103e-07,
+      "loss": 1.1009,
+      "step": 4617
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.130672986997062e-07,
+      "loss": 0.9503,
+      "step": 4618
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.116908851729238e-07,
+      "loss": 0.989,
+      "step": 4619
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.103188843147641e-07,
+      "loss": 0.9015,
+      "step": 4620
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0895129674374016e-07,
+      "loss": 0.9041,
+      "step": 4621
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0758812307637433e-07,
+      "loss": 1.0278,
+      "step": 4622
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.062293639271973e-07,
+      "loss": 1.0264,
+      "step": 4623
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.0487501990875346e-07,
+      "loss": 0.9128,
+      "step": 4624
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.035250916315934e-07,
+      "loss": 0.9832,
+      "step": 4625
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.0217957970427805e-07,
+      "loss": 0.8938,
+      "step": 4626
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.0083848473337997e-07,
+      "loss": 1.168,
+      "step": 4627
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9950180732347558e-07,
+      "loss": 0.8435,
+      "step": 4628
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9816954807715505e-07,
+      "loss": 1.1491,
+      "step": 4629
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.968417075950124e-07,
+      "loss": 0.9294,
+      "step": 4630
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.955182864756522e-07,
+      "loss": 1.039,
+      "step": 4631
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.941992853156871e-07,
+      "loss": 0.9629,
+      "step": 4632
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9288470470973487e-07,
+      "loss": 0.8957,
+      "step": 4633
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.915745452504214e-07,
+      "loss": 1.0473,
+      "step": 4634
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9026880752838205e-07,
+      "loss": 1.0696,
+      "step": 4635
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8896749213225375e-07,
+      "loss": 1.2073,
+      "step": 4636
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8767059964868384e-07,
+      "loss": 1.1135,
+      "step": 4637
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.863781306623258e-07,
+      "loss": 0.9773,
+      "step": 4638
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8509008575583465e-07,
+      "loss": 0.9164,
+      "step": 4639
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.83806465509877e-07,
+      "loss": 0.9778,
+      "step": 4640
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.8252727050312114e-07,
+      "loss": 1.2265,
+      "step": 4641
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.8125250131223903e-07,
+      "loss": 1.0637,
+      "step": 4642
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7998215851191214e-07,
+      "loss": 1.0444,
+      "step": 4643
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7871624267482345e-07,
+      "loss": 0.9814,
+      "step": 4644
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.774547543716587e-07,
+      "loss": 1.0707,
+      "step": 4645
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7619769417111188e-07,
+      "loss": 1.0365,
+      "step": 4646
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7494506263987743e-07,
+      "loss": 0.9401,
+      "step": 4647
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7369686034265476e-07,
+      "loss": 1.2595,
+      "step": 4648
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7245308784214375e-07,
+      "loss": 1.0108,
+      "step": 4649
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7121374569905257e-07,
+      "loss": 1.2021,
+      "step": 4650
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6997883447208873e-07,
+      "loss": 1.0128,
+      "step": 4651
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6874835471796025e-07,
+      "loss": 0.9444,
+      "step": 4652
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6752230699138227e-07,
+      "loss": 1.0271,
+      "step": 4653
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6630069184506936e-07,
+      "loss": 0.9095,
+      "step": 4654
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6508350982973764e-07,
+      "loss": 1.0379,
+      "step": 4655
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6387076149410376e-07,
+      "loss": 0.8244,
+      "step": 4656
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6266244738488924e-07,
+      "loss": 1.1582,
+      "step": 4657
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.614585680468117e-07,
+      "loss": 0.9357,
+      "step": 4658
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.602591240225948e-07,
+      "loss": 0.9703,
+      "step": 4659
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5906411585295823e-07,
+      "loss": 1.0958,
+      "step": 4660
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5787354407662325e-07,
+      "loss": 1.0131,
+      "step": 4661
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.566874092303139e-07,
+      "loss": 0.9575,
+      "step": 4662
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5550571184874906e-07,
+      "loss": 1.0581,
+      "step": 4663
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5432845246465266e-07,
+      "loss": 1.0942,
+      "step": 4664
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5315563160874348e-07,
+      "loss": 0.8497,
+      "step": 4665
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5198724980973967e-07,
+      "loss": 0.8311,
+      "step": 4666
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5082330759436215e-07,
+      "loss": 0.9635,
+      "step": 4667
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4966380548732674e-07,
+      "loss": 0.824,
+      "step": 4668
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4850874401134862e-07,
+      "loss": 0.9749,
+      "step": 4669
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4735812368714241e-07,
+      "loss": 0.9264,
+      "step": 4670
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.462119450334165e-07,
+      "loss": 0.9862,
+      "step": 4671
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4507020856688204e-07,
+      "loss": 1.041,
+      "step": 4672
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4393291480224502e-07,
+      "loss": 1.1229,
+      "step": 4673
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4280006425220872e-07,
+      "loss": 1.0532,
+      "step": 4674
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4167165742747458e-07,
+      "loss": 1.1156,
+      "step": 4675
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4054769483673792e-07,
+      "loss": 0.9534,
+      "step": 4676
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3942817698669343e-07,
+      "loss": 0.9365,
+      "step": 4677
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3831310438203071e-07,
+      "loss": 0.9963,
+      "step": 4678
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3720247752543658e-07,
+      "loss": 1.1204,
+      "step": 4679
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3609629691759274e-07,
+      "loss": 0.9162,
+      "step": 4680
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3499456305717584e-07,
+      "loss": 0.7906,
+      "step": 4681
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.338972764408586e-07,
+      "loss": 0.8485,
+      "step": 4682
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3280443756331084e-07,
+      "loss": 1.2288,
+      "step": 4683
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3171604691719186e-07,
+      "loss": 1.1172,
+      "step": 4684
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3063210499316247e-07,
+      "loss": 1.0713,
+      "step": 4685
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.295526122798729e-07,
+      "loss": 0.9345,
+      "step": 4686
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2847756926396947e-07,
+      "loss": 1.0492,
+      "step": 4687
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2740697643009224e-07,
+      "loss": 1.069,
+      "step": 4688
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.263408342608763e-07,
+      "loss": 1.086,
+      "step": 4689
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.2527914323694935e-07,
+      "loss": 0.9571,
+      "step": 4690
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.2422190383693079e-07,
+      "loss": 1.1092,
+      "step": 4691
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.2316911653743712e-07,
+      "loss": 0.9567,
+      "step": 4692
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.2212078181307318e-07,
+      "loss": 0.9309,
+      "step": 4693
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.2107690013644091e-07,
+      "loss": 1.171,
+      "step": 4694
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.2003747197813053e-07,
+      "loss": 0.7998,
+      "step": 4695
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1900249780672945e-07,
+      "loss": 0.9623,
+      "step": 4696
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1797197808881222e-07,
+      "loss": 0.8389,
+      "step": 4697
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1694591328894722e-07,
+      "loss": 0.9608,
+      "step": 4698
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1592430386969778e-07,
+      "loss": 0.9665,
+      "step": 4699
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1490715029161326e-07,
+      "loss": 0.9617,
+      "step": 4700
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1389445301323577e-07,
+      "loss": 1.0259,
+      "step": 4701
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1288621249110343e-07,
+      "loss": 1.0031,
+      "step": 4702
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1188242917973713e-07,
+      "loss": 0.8945,
+      "step": 4703
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1088310353165488e-07,
+      "loss": 0.9167,
+      "step": 4704
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0988823599736298e-07,
+      "loss": 0.9242,
+      "step": 4705
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.08897827025356e-07,
+      "loss": 0.7549,
+      "step": 4706
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0791187706212236e-07,
+      "loss": 1.149,
+      "step": 4707
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.069303865521365e-07,
+      "loss": 1.1042,
+      "step": 4708
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0595335593786449e-07,
+      "loss": 0.9927,
+      "step": 4709
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0498078565976289e-07,
+      "loss": 0.9856,
+      "step": 4710
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.040126761562743e-07,
+      "loss": 0.8195,
+      "step": 4711
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0304902786383409e-07,
+      "loss": 0.9343,
+      "step": 4712
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0208984121686361e-07,
+      "loss": 0.8522,
+      "step": 4713
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0113511664777476e-07,
+      "loss": 0.8351,
+      "step": 4714
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0018485458696548e-07,
+      "loss": 0.9672,
+      "step": 4715
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.923905546282642e-08,
+      "loss": 1.027,
+      "step": 4716
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.829771970172874e-08,
+      "loss": 1.293,
+      "step": 4717
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.736084772804077e-08,
+      "loss": 0.9073,
+      "step": 4718
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.642843996411134e-08,
+      "loss": 1.2303,
+      "step": 4719
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.550049683027862e-08,
+      "loss": 1.0208,
+      "step": 4720
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.457701874487135e-08,
+      "loss": 1.2482,
+      "step": 4721
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.365800612420096e-08,
+      "loss": 1.0021,
+      "step": 4722
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.274345938256824e-08,
+      "loss": 0.9343,
+      "step": 4723
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.183337893226007e-08,
+      "loss": 0.8467,
+      "step": 4724
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.092776518354939e-08,
+      "loss": 0.9806,
+      "step": 4725
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.00266185446963e-08,
+      "loss": 1.132,
+      "step": 4726
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.912993942194692e-08,
+      "loss": 1.0524,
+      "step": 4727
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.823772821953347e-08,
+      "loss": 0.9394,
+      "step": 4728
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.73499853396731e-08,
+      "loss": 0.996,
+      "step": 4729
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.64667111825701e-08,
+      "loss": 1.1593,
+      "step": 4730
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.558790614641266e-08,
+      "loss": 0.9195,
+      "step": 4731
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.471357062737606e-08,
+      "loss": 0.7653,
+      "step": 4732
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.384370501961725e-08,
+      "loss": 1.0184,
+      "step": 4733
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.297830971528364e-08,
+      "loss": 1.0217,
+      "step": 4734
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.211738510450096e-08,
+      "loss": 0.9694,
+      "step": 4735
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.12609315753854e-08,
+      "loss": 0.8418,
+      "step": 4736
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.040894951403366e-08,
+      "loss": 1.0817,
+      "step": 4737
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.956143930452743e-08,
+      "loss": 1.0594,
+      "step": 4738
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.871840132893327e-08,
+      "loss": 1.0131,
+      "step": 4739
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.787983596730275e-08,
+      "loss": 0.941,
+      "step": 4740
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.704574359766681e-08,
+      "loss": 1.0224,
+      "step": 4741
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.621612459604466e-08,
+      "loss": 0.8744,
+      "step": 4742
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.539097933643603e-08,
+      "loss": 1.0018,
+      "step": 4743
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.457030819082556e-08,
+      "loss": 1.133,
+      "step": 4744
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.375411152917955e-08,
+      "loss": 1.1121,
+      "step": 4745
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.294238971944701e-08,
+      "loss": 0.8512,
+      "step": 4746
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.213514312756076e-08,
+      "loss": 0.9784,
+      "step": 4747
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.133237211743748e-08,
+      "loss": 1.1176,
+      "step": 4748
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.053407705097216e-08,
+      "loss": 1.0379,
+      "step": 4749
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.974025828804575e-08,
+      "loss": 0.7423,
+      "step": 4750
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.895091618651873e-08,
+      "loss": 1.1176,
+      "step": 4751
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.816605110223418e-08,
+      "loss": 0.8566,
+      "step": 4752
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.738566338901909e-08,
+      "loss": 0.8822,
+      "step": 4753
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.66097533986787e-08,
+      "loss": 0.9703,
+      "step": 4754
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.583832148099988e-08,
+      "loss": 1.0903,
+      "step": 4755
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.507136798375557e-08,
+      "loss": 0.9917,
+      "step": 4756
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.430889325269251e-08,
+      "loss": 1.0903,
+      "step": 4757
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.355089763154354e-08,
+      "loss": 0.9205,
+      "step": 4758
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.279738146202196e-08,
+      "loss": 1.1939,
+      "step": 4759
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.20483450838183e-08,
+      "loss": 1.011,
+      "step": 4760
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.130378883460686e-08,
+      "loss": 0.9375,
+      "step": 4761
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.056371305004027e-08,
+      "loss": 0.9487,
+      "step": 4762
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.982811806375277e-08,
+      "loss": 0.8955,
+      "step": 4763
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.909700420735798e-08,
+      "loss": 0.8684,
+      "step": 4764
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.837037181045002e-08,
+      "loss": 0.8434,
+      "step": 4765
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.764822120060132e-08,
+      "loss": 1.2204,
+      "step": 4766
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.693055270336478e-08,
+      "loss": 1.1137,
+      "step": 4767
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.621736664227273e-08,
+      "loss": 0.9611,
+      "step": 4768
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.550866333883687e-08,
+      "loss": 1.0138,
+      "step": 4769
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.480444311254718e-08,
+      "loss": 1.1279,
+      "step": 4770
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.4104706280874166e-08,
+      "loss": 1.006,
+      "step": 4771
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.340945315926549e-08,
+      "loss": 1.0005,
+      "step": 4772
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.271868406114822e-08,
+      "loss": 1.0701,
+      "step": 4773
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.20323992979288e-08,
+      "loss": 1.0578,
+      "step": 4774
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.1350599178990875e-08,
+      "loss": 0.8568,
+      "step": 4775
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.067328401169636e-08,
+      "loss": 1.0329,
+      "step": 4776
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.0000454101386545e-08,
+      "loss": 1.1984,
+      "step": 4777
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.933210975137992e-08,
+      "loss": 1.1076,
+      "step": 4778
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.8668251262971033e-08,
+      "loss": 0.9131,
+      "step": 4779
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.800887893543826e-08,
+      "loss": 1.0437,
+      "step": 4780
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.735399306602828e-08,
+      "loss": 1.1254,
+      "step": 4781
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.67035939499727e-08,
+      "loss": 1.1217,
+      "step": 4782
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.605768188047921e-08,
+      "loss": 1.0719,
+      "step": 4783
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.5416257148729324e-08,
+      "loss": 0.8446,
+      "step": 4784
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.477932004388508e-08,
+      "loss": 0.9146,
+      "step": 4785
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.414687085308456e-08,
+      "loss": 0.8488,
+      "step": 4786
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.3518909861439695e-08,
+      "loss": 1.078,
+      "step": 4787
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.2895437352044044e-08,
+      "loss": 1.1671,
+      "step": 4788
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.2276453605965e-08,
+      "loss": 0.9095,
+      "step": 4789
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.1661958902246e-08,
+      "loss": 0.9306,
+      "step": 4790
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.105195351790769e-08,
+      "loss": 0.8666,
+      "step": 4791
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.0446437727946766e-08,
+      "loss": 1.0975,
+      "step": 4792
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.9845411805337075e-08,
+      "loss": 1.1295,
+      "step": 4793
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.9248876021024115e-08,
+      "loss": 1.0381,
+      "step": 4794
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.865683064393499e-08,
+      "loss": 0.9923,
+      "step": 4795
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.806927594096843e-08,
+      "loss": 1.1548,
+      "step": 4796
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.748621217700032e-08,
+      "loss": 0.9328,
+      "step": 4797
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.690763961488264e-08,
+      "loss": 1.1202,
+      "step": 4798
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.6333558515440073e-08,
+      "loss": 1.0433,
+      "step": 4799
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.576396913747559e-08,
+      "loss": 0.9537,
+      "step": 4800
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.519887173776493e-08,
+      "loss": 1.0154,
+      "step": 4801
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.463826657105984e-08,
+      "loss": 1.1822,
+      "step": 4802
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.4082153890088174e-08,
+      "loss": 1.1625,
+      "step": 4803
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.353053394554828e-08,
+      "loss": 0.8827,
+      "step": 4804
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.2983406986119015e-08,
+      "loss": 1.0881,
+      "step": 4805
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.2440773258448635e-08,
+      "loss": 0.9759,
+      "step": 4806
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.190263300716257e-08,
+      "loss": 0.7524,
+      "step": 4807
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.136898647486009e-08,
+      "loss": 0.9775,
+      "step": 4808
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.0839833902114315e-08,
+      "loss": 1.2051,
+      "step": 4809
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.031517552747332e-08,
+      "loss": 0.9668,
+      "step": 4810
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.979501158745679e-08,
+      "loss": 1.0348,
+      "step": 4811
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.9279342316561597e-08,
+      "loss": 1.2013,
+      "step": 4812
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.876816794725512e-08,
+      "loss": 0.9704,
+      "step": 4813
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.82614887099808e-08,
+      "loss": 0.8593,
+      "step": 4814
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.775930483315481e-08,
+      "loss": 0.951,
+      "step": 4815
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.7261616543166058e-08,
+      "loss": 1.2031,
+      "step": 4816
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.6768424064378406e-08,
+      "loss": 0.9256,
+      "step": 4817
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.6279727619127338e-08,
+      "loss": 0.9771,
+      "step": 4818
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.5795527427722177e-08,
+      "loss": 0.8893,
+      "step": 4819
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.5315823708446098e-08,
+      "loss": 0.9442,
+      "step": 4820
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.4840616677552776e-08,
+      "loss": 1.177,
+      "step": 4821
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.436990654927085e-08,
+      "loss": 1.1798,
+      "step": 4822
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.3903693535802796e-08,
+      "loss": 0.9748,
+      "step": 4823
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.3441977847320497e-08,
+      "loss": 1.0097,
+      "step": 4824
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.2984759691969673e-08,
+      "loss": 1.0105,
+      "step": 4825
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.2532039275871e-08,
+      "loss": 0.7581,
+      "step": 4826
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.2083816803113444e-08,
+      "loss": 1.1774,
+      "step": 4827
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.1640092475760933e-08,
+      "loss": 0.8835,
+      "step": 4828
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.1200866493849003e-08,
+      "loss": 1.0376,
+      "step": 4829
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0766139055383716e-08,
+      "loss": 1.023,
+      "step": 4830
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.033591035634719e-08,
+      "loss": 1.1745,
+      "step": 4831
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9910180590689833e-08,
+      "loss": 0.8797,
+      "step": 4832
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.948894995033368e-08,
+      "loss": 0.8719,
+      "step": 4833
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.907221862517572e-08,
+      "loss": 1.038,
+      "step": 4834
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.865998680308123e-08,
+      "loss": 0.8938,
+      "step": 4835
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8252254669888224e-08,
+      "loss": 1.1904,
+      "step": 4836
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.7849022409407446e-08,
+      "loss": 1.1065,
+      "step": 4837
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.745029020342015e-08,
+      "loss": 0.7901,
+      "step": 4838
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.7056058231679217e-08,
+      "loss": 0.6729,
+      "step": 4839
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.6666326671905818e-08,
+      "loss": 1.0279,
+      "step": 4840
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.6281095699798298e-08,
+      "loss": 0.9262,
+      "step": 4841
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5900365489019964e-08,
+      "loss": 0.9168,
+      "step": 4842
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5524136211209073e-08,
+      "loss": 1.0074,
+      "step": 4843
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5152408035974396e-08,
+      "loss": 1.0974,
+      "step": 4844
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4785181130893e-08,
+      "loss": 1.0856,
+      "step": 4845
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4422455661515789e-08,
+      "loss": 0.7969,
+      "step": 4846
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.406423179136196e-08,
+      "loss": 1.0079,
+      "step": 4847
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3710509681923445e-08,
+      "loss": 0.8899,
+      "step": 4848
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3361289492661577e-08,
+      "loss": 0.8229,
+      "step": 4849
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3016571381008204e-08,
+      "loss": 1.135,
+      "step": 4850
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2676355502365678e-08,
+      "loss": 1.0265,
+      "step": 4851
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2340642010105763e-08,
+      "loss": 0.8324,
+      "step": 4852
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.200943105557295e-08,
+      "loss": 0.8886,
+      "step": 4853
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.1682722788080025e-08,
+      "loss": 1.0656,
+      "step": 4854
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.1360517354911394e-08,
+      "loss": 1.1546,
+      "step": 4855
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.1042814901318644e-08,
+      "loss": 1.0796,
+      "step": 4856
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0729615570526098e-08,
+      "loss": 0.9039,
+      "step": 4857
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0420919503726368e-08,
+      "loss": 1.0888,
+      "step": 4858
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0116726840084801e-08,
+      "loss": 0.904,
+      "step": 4859
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.817037716732813e-09,
+      "loss": 1.0816,
+      "step": 4860
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.521852268773446e-09,
+      "loss": 1.0229,
+      "step": 4861
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.231170629280251e-09,
+      "loss": 1.1692,
+      "step": 4862
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.94499292929396e-09,
+      "loss": 0.9808,
+      "step": 4863
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.663319297828044e-09,
+      "loss": 0.9044,
+      "step": 4864
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.386149861862037e-09,
+      "loss": 0.8473,
+      "step": 4865
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.11348474634821e-09,
+      "loss": 0.8804,
+      "step": 4866
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.845324074207128e-09,
+      "loss": 1.0646,
+      "step": 4867
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.581667966325424e-09,
+      "loss": 0.8721,
+      "step": 4868
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.322516541565794e-09,
+      "loss": 0.9748,
+      "step": 4869
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.067869916753678e-09,
+      "loss": 0.8811,
+      "step": 4870
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.817728206687246e-09,
+      "loss": 0.8389,
+      "step": 4871
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.5720915241318515e-09,
+      "loss": 1.0244,
+      "step": 4872
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.3309599798244695e-09,
+      "loss": 1.0709,
+      "step": 4873
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.0943336824692556e-09,
+      "loss": 1.049,
+      "step": 4874
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.86221273873977e-09,
+      "loss": 1.3068,
+      "step": 4875
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.634597253277862e-09,
+      "loss": 1.0919,
+      "step": 4876
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.411487328695897e-09,
+      "loss": 0.8287,
+      "step": 4877
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.1928830655723075e-09,
+      "loss": 1.1369,
+      "step": 4878
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.978784562458261e-09,
+      "loss": 0.9869,
+      "step": 4879
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.769191915869886e-09,
+      "loss": 1.1387,
+      "step": 4880
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.564105220294934e-09,
+      "loss": 1.0444,
+      "step": 4881
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.363524568188338e-09,
+      "loss": 0.9461,
+      "step": 4882
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.167450049972211e-09,
+      "loss": 0.8991,
+      "step": 4883
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.9758817540414e-09,
+      "loss": 0.9601,
+      "step": 4884
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.788819766754603e-09,
+      "loss": 0.952,
+      "step": 4885
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.606264172443252e-09,
+      "loss": 0.9291,
+      "step": 4886
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.4282150534037383e-09,
+      "loss": 1.1345,
+      "step": 4887
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.254672489902966e-09,
+      "loss": 1.1119,
+      "step": 4888
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.0856365601750203e-09,
+      "loss": 1.138,
+      "step": 4889
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.9211073404245006e-09,
+      "loss": 0.9141,
+      "step": 4890
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.7610849048220755e-09,
+      "loss": 0.9553,
+      "step": 4891
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.605569325506707e-09,
+      "loss": 1.1938,
+      "step": 4892
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.454560672587869e-09,
+      "loss": 0.8143,
+      "step": 4893
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.3080590141411064e-09,
+      "loss": 1.2185,
+      "step": 4894
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.166064416210256e-09,
+      "loss": 1.1742,
+      "step": 4895
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.0285769428074477e-09,
+      "loss": 0.9997,
+      "step": 4896
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8955966559153217e-09,
+      "loss": 1.0956,
+      "step": 4897
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.7671236154814808e-09,
+      "loss": 1.1363,
+      "step": 4898
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.64315787942404e-09,
+      "loss": 1.0179,
+      "step": 4899
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.523699503626075e-09,
+      "loss": 0.916,
+      "step": 4900
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4087485419422842e-09,
+      "loss": 1.0157,
+      "step": 4901
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.2983050461923275e-09,
+      "loss": 1.0697,
+      "step": 4902
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.192369066167487e-09,
+      "loss": 1.0056,
+      "step": 4903
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.0909406496217857e-09,
+      "loss": 1.0678,
+      "step": 4904
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.940198422830894e-10,
+      "loss": 1.382,
+      "step": 4905
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.016066878420049e-10,
+      "loss": 0.934,
+      "step": 4906
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.137012279596513e-10,
+      "loss": 1.1835,
+      "step": 4907
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.303035022665495e-10,
+      "loss": 1.1135,
+      "step": 4908
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.514135483570716e-10,
+      "loss": 1.1104,
+      "step": 4909
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.770314017972123e-10,
+      "loss": 0.9515,
+      "step": 4910
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.071570961179273e-10,
+      "loss": 1.1433,
+      "step": 4911
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.417906628195745e-10,
+      "loss": 1.0924,
+      "step": 4912
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.8093213137080363e-10,
+      "loss": 0.8609,
+      "step": 4913
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.245815292063359e-10,
+      "loss": 0.9772,
+      "step": 4914
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.727388817302945e-10,
+      "loss": 0.9141,
+      "step": 4915
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.2540421231287413e-10,
+      "loss": 1.0775,
+      "step": 4916
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8257754229478175e-10,
+      "loss": 1.0589,
+      "step": 4917
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4425889098057533e-10,
+      "loss": 0.7837,
+      "step": 4918
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.1044827564643534e-10,
+      "loss": 1.2056,
+      "step": 4919
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 8.114571153350349e-11,
+      "loss": 1.1105,
+      "step": 4920
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 5.6351211851213285e-11,
+      "loss": 1.0876,
+      "step": 4921
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.606478777840039e-11,
+      "loss": 1.1333,
+      "step": 4922
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.028644845997185e-11,
+      "loss": 1.0439,
+      "step": 4923
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 9.01620100912659e-12,
+      "loss": 0.8314,
+      "step": 4924
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.2540505062451647e-12,
+      "loss": 1.0909,
+      "step": 4925
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0,
+      "loss": 1.0309,
+      "step": 4926
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 4926,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 500,
+  "total_flos": 6.256347214436631e+18,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}