diff --git "a/internvl2_8b_full_finetune/trainer_state.json" "b/internvl2_8b_full_finetune/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/internvl2_8b_full_finetune/trainer_state.json"
@@ -0,0 +1,28554 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9999474154703686,
+  "eval_steps": 500,
+  "global_step": 4754,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7972027972027973e-07,
+      "loss": 3.7565,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.594405594405595e-07,
+      "loss": 3.6338,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.391608391608393e-07,
+      "loss": 3.6721,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.118881118881119e-06,
+      "loss": 3.6397,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3986013986013987e-06,
+      "loss": 3.7318,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6783216783216785e-06,
+      "loss": 3.3937,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9580419580419583e-06,
+      "loss": 3.173,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.237762237762238e-06,
+      "loss": 3.0256,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.517482517482518e-06,
+      "loss": 2.2789,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7972027972027974e-06,
+      "loss": 2.1252,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0769230769230774e-06,
+      "loss": 1.8772,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.356643356643357e-06,
+      "loss": 1.5638,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.6363636363636366e-06,
+      "loss": 1.5352,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.916083916083917e-06,
+      "loss": 1.4646,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.195804195804197e-06,
+      "loss": 1.4345,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.475524475524476e-06,
+      "loss": 1.4364,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.755244755244756e-06,
+      "loss": 1.4582,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.034965034965036e-06,
+      "loss": 1.4391,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.314685314685315e-06,
+      "loss": 1.4497,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.594405594405595e-06,
+      "loss": 1.45,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.874125874125874e-06,
+      "loss": 1.4419,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.153846153846155e-06,
+      "loss": 1.4565,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.433566433566434e-06,
+      "loss": 1.4548,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.713286713286714e-06,
+      "loss": 1.3943,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.993006993006993e-06,
+      "loss": 1.4182,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.272727272727273e-06,
+      "loss": 1.4243,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.552447552447552e-06,
+      "loss": 1.4116,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.832167832167833e-06,
+      "loss": 1.4093,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.111888111888112e-06,
+      "loss": 1.4619,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.391608391608393e-06,
+      "loss": 1.4145,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.671328671328672e-06,
+      "loss": 1.3979,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.951048951048951e-06,
+      "loss": 1.4465,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.230769230769232e-06,
+      "loss": 1.3406,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.510489510489511e-06,
+      "loss": 1.4325,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.79020979020979e-06,
+      "loss": 1.4771,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0069930069930071e-05,
+      "loss": 1.3757,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.034965034965035e-05,
+      "loss": 1.4177,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.062937062937063e-05,
+      "loss": 1.3953,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0909090909090909e-05,
+      "loss": 1.3675,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.118881118881119e-05,
+      "loss": 1.3516,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1468531468531469e-05,
+      "loss": 1.3917,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1748251748251748e-05,
+      "loss": 1.337,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2027972027972027e-05,
+      "loss": 1.3242,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.230769230769231e-05,
+      "loss": 1.3318,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2587412587412589e-05,
+      "loss": 1.4071,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2867132867132868e-05,
+      "loss": 1.392,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3146853146853147e-05,
+      "loss": 1.4016,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3426573426573428e-05,
+      "loss": 1.3953,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3706293706293707e-05,
+      "loss": 1.3797,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3986013986013986e-05,
+      "loss": 1.3455,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4265734265734267e-05,
+      "loss": 1.4462,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4545454545454546e-05,
+      "loss": 1.3536,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4825174825174825e-05,
+      "loss": 1.4016,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5104895104895105e-05,
+      "loss": 1.441,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5384615384615387e-05,
+      "loss": 1.4135,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5664335664335666e-05,
+      "loss": 1.4371,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5944055944055945e-05,
+      "loss": 1.4199,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6223776223776225e-05,
+      "loss": 1.373,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6503496503496507e-05,
+      "loss": 1.4099,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6783216783216786e-05,
+      "loss": 1.3375,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7062937062937065e-05,
+      "loss": 1.3789,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7342657342657345e-05,
+      "loss": 1.4144,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7622377622377624e-05,
+      "loss": 1.3711,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7902097902097903e-05,
+      "loss": 1.4263,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8181818181818182e-05,
+      "loss": 1.3606,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8461538461538465e-05,
+      "loss": 1.3801,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8741258741258744e-05,
+      "loss": 1.414,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9020979020979023e-05,
+      "loss": 1.3674,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9300699300699302e-05,
+      "loss": 1.4363,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.958041958041958e-05,
+      "loss": 1.3421,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.986013986013986e-05,
+      "loss": 1.4186,
+      "step": 71
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0139860139860143e-05,
+      "loss": 1.7448,
+      "step": 72
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0419580419580422e-05,
+      "loss": 1.3815,
+      "step": 73
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.06993006993007e-05,
+      "loss": 1.3493,
+      "step": 74
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.097902097902098e-05,
+      "loss": 1.3953,
+      "step": 75
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.125874125874126e-05,
+      "loss": 1.4071,
+      "step": 76
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.153846153846154e-05,
+      "loss": 1.3465,
+      "step": 77
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1818181818181818e-05,
+      "loss": 1.4073,
+      "step": 78
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2097902097902097e-05,
+      "loss": 1.4047,
+      "step": 79
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.237762237762238e-05,
+      "loss": 1.3887,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.265734265734266e-05,
+      "loss": 1.3287,
+      "step": 81
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2937062937062938e-05,
+      "loss": 1.3959,
+      "step": 82
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3216783216783217e-05,
+      "loss": 1.3785,
+      "step": 83
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3496503496503496e-05,
+      "loss": 1.4571,
+      "step": 84
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3776223776223775e-05,
+      "loss": 1.3935,
+      "step": 85
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4055944055944054e-05,
+      "loss": 1.4453,
+      "step": 86
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.433566433566434e-05,
+      "loss": 1.3951,
+      "step": 87
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.461538461538462e-05,
+      "loss": 1.3483,
+      "step": 88
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.48951048951049e-05,
+      "loss": 1.3968,
+      "step": 89
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5174825174825178e-05,
+      "loss": 1.3972,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5454545454545457e-05,
+      "loss": 1.4067,
+      "step": 91
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5734265734265736e-05,
+      "loss": 1.428,
+      "step": 92
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6013986013986015e-05,
+      "loss": 1.4095,
+      "step": 93
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6293706293706294e-05,
+      "loss": 1.4353,
+      "step": 94
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6573426573426577e-05,
+      "loss": 1.2736,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6853146853146856e-05,
+      "loss": 1.4096,
+      "step": 96
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7132867132867135e-05,
+      "loss": 1.3583,
+      "step": 97
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7412587412587414e-05,
+      "loss": 1.3911,
+      "step": 98
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7692307692307694e-05,
+      "loss": 1.3625,
+      "step": 99
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7972027972027973e-05,
+      "loss": 1.3575,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8251748251748252e-05,
+      "loss": 1.2927,
+      "step": 101
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8531468531468534e-05,
+      "loss": 1.3838,
+      "step": 102
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8811188811188814e-05,
+      "loss": 1.3662,
+      "step": 103
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9090909090909093e-05,
+      "loss": 1.3649,
+      "step": 104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9370629370629372e-05,
+      "loss": 1.3942,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.965034965034965e-05,
+      "loss": 1.4093,
+      "step": 106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.993006993006993e-05,
+      "loss": 1.3986,
+      "step": 107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.020979020979021e-05,
+      "loss": 1.3586,
+      "step": 108
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.048951048951049e-05,
+      "loss": 1.4076,
+      "step": 109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0769230769230774e-05,
+      "loss": 1.3857,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.104895104895105e-05,
+      "loss": 1.3899,
+      "step": 111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.132867132867133e-05,
+      "loss": 1.3579,
+      "step": 112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.160839160839161e-05,
+      "loss": 1.3288,
+      "step": 113
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.188811188811189e-05,
+      "loss": 1.3219,
+      "step": 114
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.216783216783217e-05,
+      "loss": 1.4371,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.244755244755245e-05,
+      "loss": 1.4103,
+      "step": 116
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.272727272727273e-05,
+      "loss": 1.3679,
+      "step": 117
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3006993006993014e-05,
+      "loss": 1.4001,
+      "step": 118
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.328671328671329e-05,
+      "loss": 1.3716,
+      "step": 119
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.356643356643357e-05,
+      "loss": 1.4149,
+      "step": 120
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.384615384615385e-05,
+      "loss": 1.3703,
+      "step": 121
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.412587412587413e-05,
+      "loss": 1.4115,
+      "step": 122
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.440559440559441e-05,
+      "loss": 1.4308,
+      "step": 123
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.468531468531469e-05,
+      "loss": 1.4214,
+      "step": 124
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.496503496503497e-05,
+      "loss": 1.3922,
+      "step": 125
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.524475524475525e-05,
+      "loss": 2.2952,
+      "step": 126
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.552447552447553e-05,
+      "loss": 1.4082,
+      "step": 127
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.5804195804195806e-05,
+      "loss": 1.4501,
+      "step": 128
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.608391608391609e-05,
+      "loss": 1.4269,
+      "step": 129
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.6363636363636364e-05,
+      "loss": 1.3986,
+      "step": 130
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.664335664335665e-05,
+      "loss": 1.429,
+      "step": 131
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.692307692307693e-05,
+      "loss": 1.3727,
+      "step": 132
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.7202797202797205e-05,
+      "loss": 1.3966,
+      "step": 133
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.748251748251749e-05,
+      "loss": 1.4479,
+      "step": 134
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.776223776223776e-05,
+      "loss": 1.2996,
+      "step": 135
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.8041958041958046e-05,
+      "loss": 1.3981,
+      "step": 136
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.832167832167832e-05,
+      "loss": 1.4238,
+      "step": 137
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.8601398601398604e-05,
+      "loss": 1.3833,
+      "step": 138
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.888111888111888e-05,
+      "loss": 1.4186,
+      "step": 139
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.916083916083916e-05,
+      "loss": 1.3688,
+      "step": 140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.9440559440559445e-05,
+      "loss": 1.3687,
+      "step": 141
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.972027972027972e-05,
+      "loss": 1.3736,
+      "step": 142
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4e-05,
+      "loss": 1.3654,
+      "step": 143
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999999535795353e-05,
+      "loss": 1.3127,
+      "step": 144
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.9999981431816256e-05,
+      "loss": 1.3956,
+      "step": 145
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999995822159466e-05,
+      "loss": 1.3325,
+      "step": 146
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.9999925727299505e-05,
+      "loss": 1.3311,
+      "step": 147
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999988394894588e-05,
+      "loss": 1.371,
+      "step": 148
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999983288655318e-05,
+      "loss": 1.3063,
+      "step": 149
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.9999772540145104e-05,
+      "loss": 1.4212,
+      "step": 150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999970290974967e-05,
+      "loss": 1.3853,
+      "step": 151
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999962399539919e-05,
+      "loss": 1.3556,
+      "step": 152
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.9999535797130304e-05,
+      "loss": 1.3849,
+      "step": 153
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999943831498395e-05,
+      "loss": 1.3933,
+      "step": 154
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.9999331549005394e-05,
+      "loss": 1.3331,
+      "step": 155
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999921549924418e-05,
+      "loss": 1.3488,
+      "step": 156
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999909016575419e-05,
+      "loss": 1.4472,
+      "step": 157
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.99989555485936e-05,
+      "loss": 1.4994,
+      "step": 158
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.99988116478249e-05,
+      "loss": 1.4134,
+      "step": 159
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999865846351489e-05,
+      "loss": 1.3904,
+      "step": 160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.9998495995734677e-05,
+      "loss": 1.3941,
+      "step": 161
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999832424455968e-05,
+      "loss": 1.3369,
+      "step": 162
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999814321006963e-05,
+      "loss": 1.3989,
+      "step": 163
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999795289234856e-05,
+      "loss": 1.3794,
+      "step": 164
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.9997753291484816e-05,
+      "loss": 1.3702,
+      "step": 165
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999754440757105e-05,
+      "loss": 1.322,
+      "step": 166
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999732624070424e-05,
+      "loss": 1.4168,
+      "step": 167
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999709879098565e-05,
+      "loss": 1.3924,
+      "step": 168
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999686205852087e-05,
+      "loss": 1.3191,
+      "step": 169
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999661604341978e-05,
+      "loss": 1.398,
+      "step": 170
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.99963607457966e-05,
+      "loss": 1.3207,
+      "step": 171
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999609616576982e-05,
+      "loss": 1.3763,
+      "step": 172
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9995822303462273e-05,
+      "loss": 1.3931,
+      "step": 173
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9995539159001074e-05,
+      "loss": 1.382,
+      "step": 174
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999524673251768e-05,
+      "loss": 1.4525,
+      "step": 175
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999494502414783e-05,
+      "loss": 1.4071,
+      "step": 176
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999463403403156e-05,
+      "loss": 1.4079,
+      "step": 177
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999431376231326e-05,
+      "loss": 1.44,
+      "step": 178
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9993984209141576e-05,
+      "loss": 1.4475,
+      "step": 179
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999364537466951e-05,
+      "loss": 1.4105,
+      "step": 180
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999329725905434e-05,
+      "loss": 1.3631,
+      "step": 181
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999293986245766e-05,
+      "loss": 1.3998,
+      "step": 182
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9992573185045386e-05,
+      "loss": 1.3647,
+      "step": 183
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9992197226987725e-05,
+      "loss": 1.4261,
+      "step": 184
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999181198845919e-05,
+      "loss": 1.3835,
+      "step": 185
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999141746963862e-05,
+      "loss": 1.3491,
+      "step": 186
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999101367070916e-05,
+      "loss": 1.3868,
+      "step": 187
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9990600591858244e-05,
+      "loss": 1.3502,
+      "step": 188
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999017823327762e-05,
+      "loss": 1.3767,
+      "step": 189
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9989746595163364e-05,
+      "loss": 1.4154,
+      "step": 190
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998930567771583e-05,
+      "loss": 1.3643,
+      "step": 191
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998885548113971e-05,
+      "loss": 1.427,
+      "step": 192
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998839600564398e-05,
+      "loss": 1.3278,
+      "step": 193
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998792725144192e-05,
+      "loss": 1.364,
+      "step": 194
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9987449218751134e-05,
+      "loss": 1.3331,
+      "step": 195
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998696190779354e-05,
+      "loss": 1.3721,
+      "step": 196
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9986465318795336e-05,
+      "loss": 1.4242,
+      "step": 197
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998595945198705e-05,
+      "loss": 1.3941,
+      "step": 198
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9985444307603497e-05,
+      "loss": 1.398,
+      "step": 199
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998491988588381e-05,
+      "loss": 1.3835,
+      "step": 200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998438618707144e-05,
+      "loss": 1.383,
+      "step": 201
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9983843211414124e-05,
+      "loss": 1.4021,
+      "step": 202
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9983290959163914e-05,
+      "loss": 1.3831,
+      "step": 203
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998272943057717e-05,
+      "loss": 1.3201,
+      "step": 204
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998215862591455e-05,
+      "loss": 1.3822,
+      "step": 205
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998157854544104e-05,
+      "loss": 1.3514,
+      "step": 206
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.99809891894259e-05,
+      "loss": 1.3992,
+      "step": 207
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998039055814272e-05,
+      "loss": 1.3909,
+      "step": 208
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9979782651869384e-05,
+      "loss": 1.3097,
+      "step": 209
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.997916547088808e-05,
+      "loss": 1.3593,
+      "step": 210
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.997853901548532e-05,
+      "loss": 1.3284,
+      "step": 211
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9977903285951896e-05,
+      "loss": 1.3946,
+      "step": 212
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9977258282582916e-05,
+      "loss": 1.3739,
+      "step": 213
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.99766040056778e-05,
+      "loss": 1.371,
+      "step": 214
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.997594045554027e-05,
+      "loss": 1.3733,
+      "step": 215
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9975267632478336e-05,
+      "loss": 1.3776,
+      "step": 216
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.997458553680434e-05,
+      "loss": 1.3519,
+      "step": 217
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.99738941688349e-05,
+      "loss": 1.3254,
+      "step": 218
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.997319352889096e-05,
+      "loss": 1.3786,
+      "step": 219
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.997248361729777e-05,
+      "loss": 1.3308,
+      "step": 220
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.997176443438485e-05,
+      "loss": 1.3666,
+      "step": 221
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.997103598048607e-05,
+      "loss": 1.3565,
+      "step": 222
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9970298255939564e-05,
+      "loss": 1.342,
+      "step": 223
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9969551261087806e-05,
+      "loss": 1.4034,
+      "step": 224
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.996879499627754e-05,
+      "loss": 1.3279,
+      "step": 225
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.996802946185984e-05,
+      "loss": 1.3397,
+      "step": 226
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9967254658190055e-05,
+      "loss": 1.4121,
+      "step": 227
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.996647058562786e-05,
+      "loss": 1.3868,
+      "step": 228
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9965677244537226e-05,
+      "loss": 1.2964,
+      "step": 229
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9964874635286436e-05,
+      "loss": 1.3671,
+      "step": 230
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.996406275824804e-05,
+      "loss": 1.3471,
+      "step": 231
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.996324161379894e-05,
+      "loss": 1.309,
+      "step": 232
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9962411202320296e-05,
+      "loss": 1.382,
+      "step": 233
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.99615715241976e-05,
+      "loss": 1.3367,
+      "step": 234
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.996072257982064e-05,
+      "loss": 1.3915,
+      "step": 235
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9959864369583485e-05,
+      "loss": 1.423,
+      "step": 236
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9958996893884525e-05,
+      "loss": 1.3617,
+      "step": 237
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9958120153126454e-05,
+      "loss": 1.3036,
+      "step": 238
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.995723414771625e-05,
+      "loss": 1.3381,
+      "step": 239
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9956338878065205e-05,
+      "loss": 1.3624,
+      "step": 240
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.99554343445889e-05,
+      "loss": 1.4119,
+      "step": 241
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.995452054770724e-05,
+      "loss": 1.391,
+      "step": 242
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.99535974878444e-05,
+      "loss": 1.3295,
+      "step": 243
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.995266516542887e-05,
+      "loss": 1.3407,
+      "step": 244
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.995172358089344e-05,
+      "loss": 1.3814,
+      "step": 245
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.995077273467521e-05,
+      "loss": 1.3753,
+      "step": 246
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.994981262721555e-05,
+      "loss": 1.337,
+      "step": 247
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9948843258960154e-05,
+      "loss": 1.3938,
+      "step": 248
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9947864630359005e-05,
+      "loss": 1.3571,
+      "step": 249
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.994687674186638e-05,
+      "loss": 1.3662,
+      "step": 250
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9945879593940874e-05,
+      "loss": 1.3636,
+      "step": 251
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.994487318704536e-05,
+      "loss": 1.3692,
+      "step": 252
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.994385752164703e-05,
+      "loss": 1.413,
+      "step": 253
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9942832598217345e-05,
+      "loss": 1.4184,
+      "step": 254
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9941798417232084e-05,
+      "loss": 1.3372,
+      "step": 255
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9940754979171317e-05,
+      "loss": 1.3661,
+      "step": 256
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9939702284519416e-05,
+      "loss": 1.393,
+      "step": 257
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9938640333765046e-05,
+      "loss": 1.4067,
+      "step": 258
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.993756912740117e-05,
+      "loss": 1.358,
+      "step": 259
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9936488665925045e-05,
+      "loss": 1.3745,
+      "step": 260
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.993539894983823e-05,
+      "loss": 1.3749,
+      "step": 261
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.993429997964657e-05,
+      "loss": 1.3752,
+      "step": 262
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.993319175586021e-05,
+      "loss": 1.3392,
+      "step": 263
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9932074278993604e-05,
+      "loss": 1.4217,
+      "step": 264
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.993094754956549e-05,
+      "loss": 1.4214,
+      "step": 265
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.992981156809889e-05,
+      "loss": 1.3625,
+      "step": 266
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9928666335121135e-05,
+      "loss": 1.4306,
+      "step": 267
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.992751185116385e-05,
+      "loss": 1.3936,
+      "step": 268
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.992634811676296e-05,
+      "loss": 1.3801,
+      "step": 269
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.992517513245865e-05,
+      "loss": 1.3638,
+      "step": 270
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.992399289879546e-05,
+      "loss": 1.4009,
+      "step": 271
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.992280141632216e-05,
+      "loss": 1.3206,
+      "step": 272
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9921600685591856e-05,
+      "loss": 1.3925,
+      "step": 273
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9920390707161927e-05,
+      "loss": 1.4102,
+      "step": 274
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9919171481594056e-05,
+      "loss": 1.348,
+      "step": 275
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9917943009454206e-05,
+      "loss": 1.3055,
+      "step": 276
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9916705291312646e-05,
+      "loss": 1.3889,
+      "step": 277
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.991545832774393e-05,
+      "loss": 1.3769,
+      "step": 278
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9914202119326895e-05,
+      "loss": 1.3341,
+      "step": 279
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.991293666664469e-05,
+      "loss": 1.2588,
+      "step": 280
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.991166197028474e-05,
+      "loss": 1.3687,
+      "step": 281
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9910378030838765e-05,
+      "loss": 1.3817,
+      "step": 282
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.990908484890277e-05,
+      "loss": 1.3244,
+      "step": 283
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.990778242507707e-05,
+      "loss": 1.3482,
+      "step": 284
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.990647075996624e-05,
+      "loss": 1.3568,
+      "step": 285
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9905149854179174e-05,
+      "loss": 1.3464,
+      "step": 286
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.990381970832903e-05,
+      "loss": 1.3846,
+      "step": 287
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9902480323033285e-05,
+      "loss": 1.3303,
+      "step": 288
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.990113169891367e-05,
+      "loss": 1.3757,
+      "step": 289
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.989977383659624e-05,
+      "loss": 1.3703,
+      "step": 290
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.989840673671131e-05,
+      "loss": 1.3292,
+      "step": 291
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.989703039989349e-05,
+      "loss": 1.3497,
+      "step": 292
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.989564482678168e-05,
+      "loss": 1.3751,
+      "step": 293
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.989425001801909e-05,
+      "loss": 1.384,
+      "step": 294
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9892845974253184e-05,
+      "loss": 1.3478,
+      "step": 295
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.989143269613572e-05,
+      "loss": 1.3822,
+      "step": 296
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.989001018432276e-05,
+      "loss": 1.3131,
+      "step": 297
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.988857843947463e-05,
+      "loss": 1.3686,
+      "step": 298
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.988713746225596e-05,
+      "loss": 1.4093,
+      "step": 299
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.988568725333565e-05,
+      "loss": 1.4109,
+      "step": 300
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.98842278133869e-05,
+      "loss": 1.3869,
+      "step": 301
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9882759143087194e-05,
+      "loss": 1.3923,
+      "step": 302
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9881281243118285e-05,
+      "loss": 1.3905,
+      "step": 303
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.987979411416623e-05,
+      "loss": 1.3664,
+      "step": 304
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.987829775692135e-05,
+      "loss": 1.3708,
+      "step": 305
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.987679217207827e-05,
+      "loss": 1.2989,
+      "step": 306
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.987527736033589e-05,
+      "loss": 1.3313,
+      "step": 307
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.987375332239739e-05,
+      "loss": 1.342,
+      "step": 308
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9872220058970226e-05,
+      "loss": 1.3617,
+      "step": 309
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9870677570766167e-05,
+      "loss": 1.3414,
+      "step": 310
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.986912585850123e-05,
+      "loss": 1.346,
+      "step": 311
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9867564922895724e-05,
+      "loss": 1.3326,
+      "step": 312
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.986599476467425e-05,
+      "loss": 1.3472,
+      "step": 313
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9864415384565675e-05,
+      "loss": 1.3789,
+      "step": 314
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.986282678330316e-05,
+      "loss": 1.3781,
+      "step": 315
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9861228961624146e-05,
+      "loss": 1.2815,
+      "step": 316
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.985962192027034e-05,
+      "loss": 1.3893,
+      "step": 317
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.985800565998775e-05,
+      "loss": 1.3448,
+      "step": 318
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9856380181526634e-05,
+      "loss": 1.3941,
+      "step": 319
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9854745485641556e-05,
+      "loss": 1.388,
+      "step": 320
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.985310157309135e-05,
+      "loss": 1.3739,
+      "step": 321
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.985144844463913e-05,
+      "loss": 1.3588,
+      "step": 322
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9849786101052285e-05,
+      "loss": 1.3605,
+      "step": 323
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.984811454310248e-05,
+      "loss": 1.4181,
+      "step": 324
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9846433771565655e-05,
+      "loss": 1.3938,
+      "step": 325
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9844743787222046e-05,
+      "loss": 1.3191,
+      "step": 326
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.984304459085614e-05,
+      "loss": 1.3999,
+      "step": 327
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.984133618325671e-05,
+      "loss": 1.3927,
+      "step": 328
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.983961856521682e-05,
+      "loss": 1.4153,
+      "step": 329
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.983789173753378e-05,
+      "loss": 1.3608,
+      "step": 330
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.983615570100921e-05,
+      "loss": 1.3745,
+      "step": 331
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9834410456448966e-05,
+      "loss": 1.3283,
+      "step": 332
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.983265600466321e-05,
+      "loss": 1.3819,
+      "step": 333
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.983089234646637e-05,
+      "loss": 1.3575,
+      "step": 334
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9829119482677144e-05,
+      "loss": 1.3852,
+      "step": 335
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9827337414118486e-05,
+      "loss": 1.3329,
+      "step": 336
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.982554614161766e-05,
+      "loss": 1.3351,
+      "step": 337
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9823745666006176e-05,
+      "loss": 1.3919,
+      "step": 338
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.982193598811983e-05,
+      "loss": 1.3631,
+      "step": 339
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9820117108798666e-05,
+      "loss": 1.3529,
+      "step": 340
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.981828902888704e-05,
+      "loss": 1.3448,
+      "step": 341
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.981645174923353e-05,
+      "loss": 1.3787,
+      "step": 342
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9814605270691025e-05,
+      "loss": 1.3678,
+      "step": 343
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.981274959411667e-05,
+      "loss": 1.3527,
+      "step": 344
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9810884720371874e-05,
+      "loss": 1.3531,
+      "step": 345
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.980901065032232e-05,
+      "loss": 1.3508,
+      "step": 346
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9807127384837955e-05,
+      "loss": 1.3343,
+      "step": 347
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.980523492479301e-05,
+      "loss": 1.374,
+      "step": 348
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.980333327106596e-05,
+      "loss": 1.326,
+      "step": 349
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.980142242453958e-05,
+      "loss": 1.326,
+      "step": 350
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.979950238610088e-05,
+      "loss": 1.2934,
+      "step": 351
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9797573156641165e-05,
+      "loss": 1.3524,
+      "step": 352
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9795634737055974e-05,
+      "loss": 1.3736,
+      "step": 353
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.979368712824514e-05,
+      "loss": 1.3553,
+      "step": 354
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.979173033111275e-05,
+      "loss": 1.337,
+      "step": 355
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9789764346567154e-05,
+      "loss": 1.3571,
+      "step": 356
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.978778917552099e-05,
+      "loss": 1.3803,
+      "step": 357
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9785804818891117e-05,
+      "loss": 1.3336,
+      "step": 358
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.978381127759869e-05,
+      "loss": 1.3586,
+      "step": 359
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9781808552569134e-05,
+      "loss": 1.3318,
+      "step": 360
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.97797966447321e-05,
+      "loss": 1.4023,
+      "step": 361
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.977777555502155e-05,
+      "loss": 1.3615,
+      "step": 362
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.977574528437567e-05,
+      "loss": 1.3628,
+      "step": 363
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.977370583373692e-05,
+      "loss": 1.3542,
+      "step": 364
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9771657204052026e-05,
+      "loss": 1.3581,
+      "step": 365
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.976959939627196e-05,
+      "loss": 1.3722,
+      "step": 366
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9767532411351985e-05,
+      "loss": 1.3393,
+      "step": 367
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.97654562502516e-05,
+      "loss": 1.4009,
+      "step": 368
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9763370913934554e-05,
+      "loss": 1.3111,
+      "step": 369
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.976127640336889e-05,
+      "loss": 1.3528,
+      "step": 370
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9759172719526876e-05,
+      "loss": 1.3229,
+      "step": 371
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.975705986338505e-05,
+      "loss": 1.3766,
+      "step": 372
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9754937835924214e-05,
+      "loss": 1.3366,
+      "step": 373
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9752806638129435e-05,
+      "loss": 1.2942,
+      "step": 374
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.975066627099e-05,
+      "loss": 1.3489,
+      "step": 375
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.974851673549951e-05,
+      "loss": 1.318,
+      "step": 376
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9746358032655764e-05,
+      "loss": 1.3541,
+      "step": 377
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.974419016346084e-05,
+      "loss": 1.3754,
+      "step": 378
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.974201312892109e-05,
+      "loss": 1.379,
+      "step": 379
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9739826930047095e-05,
+      "loss": 1.3513,
+      "step": 380
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9737631567853695e-05,
+      "loss": 1.3605,
+      "step": 381
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9735427043359996e-05,
+      "loss": 1.3699,
+      "step": 382
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.973321335758934e-05,
+      "loss": 1.3238,
+      "step": 383
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.973099051156934e-05,
+      "loss": 1.3938,
+      "step": 384
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.972875850633184e-05,
+      "loss": 1.3824,
+      "step": 385
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9726517342912954e-05,
+      "loss": 1.3792,
+      "step": 386
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.972426702235304e-05,
+      "loss": 1.3718,
+      "step": 387
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.972200754569671e-05,
+      "loss": 1.3514,
+      "step": 388
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9719738913992815e-05,
+      "loss": 1.3538,
+      "step": 389
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.971746112829447e-05,
+      "loss": 1.3107,
+      "step": 390
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9715174189659036e-05,
+      "loss": 1.329,
+      "step": 391
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.971287809914811e-05,
+      "loss": 1.3611,
+      "step": 392
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.971057285782757e-05,
+      "loss": 1.3872,
+      "step": 393
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.970825846676749e-05,
+      "loss": 1.3523,
+      "step": 394
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.970593492704225e-05,
+      "loss": 1.3117,
+      "step": 395
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9703602239730425e-05,
+      "loss": 1.3709,
+      "step": 396
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9701260405914874e-05,
+      "loss": 1.3922,
+      "step": 397
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9698909426682674e-05,
+      "loss": 1.3415,
+      "step": 398
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9696549303125176e-05,
+      "loss": 1.3688,
+      "step": 399
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.969418003633795e-05,
+      "loss": 1.3396,
+      "step": 400
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.969180162742082e-05,
+      "loss": 1.3672,
+      "step": 401
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9689414077477865e-05,
+      "loss": 1.3191,
+      "step": 402
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.968701738761739e-05,
+      "loss": 1.2855,
+      "step": 403
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.968461155895194e-05,
+      "loss": 1.3558,
+      "step": 404
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9682196592598324e-05,
+      "loss": 1.3334,
+      "step": 405
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.967977248967758e-05,
+      "loss": 1.3646,
+      "step": 406
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.967733925131498e-05,
+      "loss": 1.3379,
+      "step": 407
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9674896878640054e-05,
+      "loss": 1.3785,
+      "step": 408
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9672445372786565e-05,
+      "loss": 1.287,
+      "step": 409
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.96699847348925e-05,
+      "loss": 1.3839,
+      "step": 410
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.966751496610011e-05,
+      "loss": 1.3697,
+      "step": 411
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.966503606755586e-05,
+      "loss": 1.3358,
+      "step": 412
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9662548040410485e-05,
+      "loss": 1.3074,
+      "step": 413
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9660050885818925e-05,
+      "loss": 1.3352,
+      "step": 414
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.965754460494037e-05,
+      "loss": 1.2467,
+      "step": 415
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9655029198938256e-05,
+      "loss": 1.3878,
+      "step": 416
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.965250466898024e-05,
+      "loss": 1.3401,
+      "step": 417
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.964997101623823e-05,
+      "loss": 1.3288,
+      "step": 418
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.964742824188834e-05,
+      "loss": 1.2528,
+      "step": 419
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9644876347110956e-05,
+      "loss": 1.3386,
+      "step": 420
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.964231533309067e-05,
+      "loss": 1.3433,
+      "step": 421
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.963974520101632e-05,
+      "loss": 1.3288,
+      "step": 422
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.963716595208098e-05,
+      "loss": 1.3128,
+      "step": 423
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.963457758748193e-05,
+      "loss": 1.3677,
+      "step": 424
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.963198010842073e-05,
+      "loss": 1.3343,
+      "step": 425
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9629373516103114e-05,
+      "loss": 1.325,
+      "step": 426
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9626757811739084e-05,
+      "loss": 1.3301,
+      "step": 427
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.962413299654286e-05,
+      "loss": 1.3561,
+      "step": 428
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.962149907173291e-05,
+      "loss": 1.3344,
+      "step": 429
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.961885603853189e-05,
+      "loss": 1.3034,
+      "step": 430
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9616203898166724e-05,
+      "loss": 1.3253,
+      "step": 431
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.961354265186854e-05,
+      "loss": 1.3421,
+      "step": 432
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9610872300872704e-05,
+      "loss": 1.3792,
+      "step": 433
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.96081928464188e-05,
+      "loss": 1.4123,
+      "step": 434
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.960550428975066e-05,
+      "loss": 1.3507,
+      "step": 435
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9602806632116304e-05,
+      "loss": 1.362,
+      "step": 436
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.960009987476801e-05,
+      "loss": 1.3853,
+      "step": 437
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.959738401896227e-05,
+      "loss": 1.3272,
+      "step": 438
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9594659065959774e-05,
+      "loss": 1.3299,
+      "step": 439
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.959192501702548e-05,
+      "loss": 1.3923,
+      "step": 440
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.958918187342855e-05,
+      "loss": 1.3553,
+      "step": 441
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9586429636442346e-05,
+      "loss": 1.3334,
+      "step": 442
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.958366830734448e-05,
+      "loss": 1.3455,
+      "step": 443
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.958089788741677e-05,
+      "loss": 1.3692,
+      "step": 444
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.957811837794526e-05,
+      "loss": 1.3555,
+      "step": 445
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9575329780220215e-05,
+      "loss": 1.2919,
+      "step": 446
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.957253209553611e-05,
+      "loss": 1.3158,
+      "step": 447
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.956972532519164e-05,
+      "loss": 1.2939,
+      "step": 448
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.956690947048972e-05,
+      "loss": 1.392,
+      "step": 449
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9564084532737495e-05,
+      "loss": 1.3587,
+      "step": 450
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9561250513246306e-05,
+      "loss": 1.3468,
+      "step": 451
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.955840741333171e-05,
+      "loss": 1.3749,
+      "step": 452
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9555555234313506e-05,
+      "loss": 1.3428,
+      "step": 453
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9552693977515675e-05,
+      "loss": 1.359,
+      "step": 454
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9549823644266434e-05,
+      "loss": 1.4116,
+      "step": 455
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9546944235898194e-05,
+      "loss": 1.2855,
+      "step": 456
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.954405575374759e-05,
+      "loss": 1.3296,
+      "step": 457
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.954115819915549e-05,
+      "loss": 1.3525,
+      "step": 458
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9538251573466926e-05,
+      "loss": 1.3051,
+      "step": 459
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9535335878031185e-05,
+      "loss": 1.3067,
+      "step": 460
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.953241111420174e-05,
+      "loss": 1.3688,
+      "step": 461
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9529477283336274e-05,
+      "loss": 1.3524,
+      "step": 462
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9526534386796696e-05,
+      "loss": 1.3613,
+      "step": 463
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.95235824259491e-05,
+      "loss": 1.3574,
+      "step": 464
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.952062140216381e-05,
+      "loss": 1.3534,
+      "step": 465
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.951765131681535e-05,
+      "loss": 1.3318,
+      "step": 466
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9514672171282435e-05,
+      "loss": 1.3822,
+      "step": 467
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.951168396694801e-05,
+      "loss": 1.308,
+      "step": 468
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9508686705199196e-05,
+      "loss": 1.3783,
+      "step": 469
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.950568038742736e-05,
+      "loss": 1.3454,
+      "step": 470
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.950266501502803e-05,
+      "loss": 1.3656,
+      "step": 471
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9499640589400964e-05,
+      "loss": 1.3865,
+      "step": 472
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.949660711195011e-05,
+      "loss": 1.3157,
+      "step": 473
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.949356458408363e-05,
+      "loss": 1.3184,
+      "step": 474
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9490513007213874e-05,
+      "loss": 1.373,
+      "step": 475
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.94874523827574e-05,
+      "loss": 1.2863,
+      "step": 476
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9484382712134956e-05,
+      "loss": 1.3428,
+      "step": 477
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9481303996771505e-05,
+      "loss": 1.3618,
+      "step": 478
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9478216238096206e-05,
+      "loss": 1.3291,
+      "step": 479
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.94751194375424e-05,
+      "loss": 1.293,
+      "step": 480
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9472013596547646e-05,
+      "loss": 1.3879,
+      "step": 481
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.946889871655368e-05,
+      "loss": 1.3692,
+      "step": 482
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.946577479900645e-05,
+      "loss": 1.3788,
+      "step": 483
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9462641845356096e-05,
+      "loss": 1.335,
+      "step": 484
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.945949985705694e-05,
+      "loss": 1.3058,
+      "step": 485
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.945634883556752e-05,
+      "loss": 1.3531,
+      "step": 486
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.945318878235054e-05,
+      "loss": 1.2844,
+      "step": 487
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.945001969887293e-05,
+      "loss": 1.3488,
+      "step": 488
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.944684158660577e-05,
+      "loss": 1.3626,
+      "step": 489
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.944365444702437e-05,
+      "loss": 1.3428,
+      "step": 490
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.944045828160822e-05,
+      "loss": 1.3372,
+      "step": 491
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.943725309184098e-05,
+      "loss": 1.3614,
+      "step": 492
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.943403887921052e-05,
+      "loss": 1.3507,
+      "step": 493
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.94308156452089e-05,
+      "loss": 1.3577,
+      "step": 494
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9427583391332354e-05,
+      "loss": 1.3446,
+      "step": 495
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.94243421190813e-05,
+      "loss": 1.3446,
+      "step": 496
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9421091829960364e-05,
+      "loss": 1.3453,
+      "step": 497
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9417832525478344e-05,
+      "loss": 1.3483,
+      "step": 498
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.941456420714822e-05,
+      "loss": 1.3611,
+      "step": 499
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.941128687648717e-05,
+      "loss": 1.332,
+      "step": 500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.940800053501653e-05,
+      "loss": 1.3141,
+      "step": 501
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9404705184261846e-05,
+      "loss": 1.3314,
+      "step": 502
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9401400825752835e-05,
+      "loss": 1.3375,
+      "step": 503
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.939808746102339e-05,
+      "loss": 1.3588,
+      "step": 504
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9394765091611596e-05,
+      "loss": 1.3301,
+      "step": 505
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.939143371905971e-05,
+      "loss": 1.323,
+      "step": 506
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.938809334491417e-05,
+      "loss": 1.3438,
+      "step": 507
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9384743970725596e-05,
+      "loss": 1.3791,
+      "step": 508
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.938138559804878e-05,
+      "loss": 1.3477,
+      "step": 509
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9378018228442696e-05,
+      "loss": 1.3494,
+      "step": 510
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.937464186347049e-05,
+      "loss": 1.2981,
+      "step": 511
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9371256504699486e-05,
+      "loss": 1.3292,
+      "step": 512
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.936786215370119e-05,
+      "loss": 1.3768,
+      "step": 513
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.936445881205127e-05,
+      "loss": 1.3215,
+      "step": 514
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.936104648132957e-05,
+      "loss": 1.3193,
+      "step": 515
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.935762516312012e-05,
+      "loss": 1.3531,
+      "step": 516
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9354194859011105e-05,
+      "loss": 1.3579,
+      "step": 517
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.935075557059488e-05,
+      "loss": 1.3258,
+      "step": 518
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.934730729946799e-05,
+      "loss": 1.3648,
+      "step": 519
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9343850047231144e-05,
+      "loss": 1.2782,
+      "step": 520
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9340383815489204e-05,
+      "loss": 1.3193,
+      "step": 521
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.933690860585121e-05,
+      "loss": 1.4034,
+      "step": 522
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.933342441993037e-05,
+      "loss": 1.3484,
+      "step": 523
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.932993125934407e-05,
+      "loss": 1.3101,
+      "step": 524
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.932642912571385e-05,
+      "loss": 1.3515,
+      "step": 525
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.932291802066539e-05,
+      "loss": 1.3277,
+      "step": 526
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.93193979458286e-05,
+      "loss": 1.3594,
+      "step": 527
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.93158689028375e-05,
+      "loss": 1.329,
+      "step": 528
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.931233089333027e-05,
+      "loss": 1.3146,
+      "step": 529
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9308783918949296e-05,
+      "loss": 1.3698,
+      "step": 530
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9305227981341085e-05,
+      "loss": 1.349,
+      "step": 531
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.930166308215633e-05,
+      "loss": 1.3225,
+      "step": 532
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.929808922304987e-05,
+      "loss": 1.3138,
+      "step": 533
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.92945064056807e-05,
+      "loss": 1.3343,
+      "step": 534
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.929091463171199e-05,
+      "loss": 1.3186,
+      "step": 535
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.928731390281105e-05,
+      "loss": 1.3135,
+      "step": 536
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.928370422064936e-05,
+      "loss": 1.3425,
+      "step": 537
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.928008558690255e-05,
+      "loss": 1.3065,
+      "step": 538
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.927645800325041e-05,
+      "loss": 1.3098,
+      "step": 539
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.927282147137688e-05,
+      "loss": 1.3559,
+      "step": 540
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9269175992970055e-05,
+      "loss": 1.334,
+      "step": 541
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9265521569722176e-05,
+      "loss": 1.3478,
+      "step": 542
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.926185820332965e-05,
+      "loss": 1.3672,
+      "step": 543
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9258185895493026e-05,
+      "loss": 1.3822,
+      "step": 544
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.925450464791701e-05,
+      "loss": 1.3705,
+      "step": 545
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.925081446231045e-05,
+      "loss": 1.3478,
+      "step": 546
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.924711534038635e-05,
+      "loss": 1.3549,
+      "step": 547
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9243407283861866e-05,
+      "loss": 1.2949,
+      "step": 548
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.923969029445828e-05,
+      "loss": 1.3545,
+      "step": 549
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.923596437390105e-05,
+      "loss": 1.355,
+      "step": 550
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.923222952391975e-05,
+      "loss": 1.2885,
+      "step": 551
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9228485746248134e-05,
+      "loss": 1.3509,
+      "step": 552
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.922473304262406e-05,
+      "loss": 1.3077,
+      "step": 553
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.922097141478957e-05,
+      "loss": 1.3247,
+      "step": 554
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.921720086449082e-05,
+      "loss": 1.3522,
+      "step": 555
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.921342139347811e-05,
+      "loss": 1.354,
+      "step": 556
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.92096330035059e-05,
+      "loss": 1.3082,
+      "step": 557
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9205835696332775e-05,
+      "loss": 1.358,
+      "step": 558
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.920202947372146e-05,
+      "loss": 1.3389,
+      "step": 559
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.919821433743882e-05,
+      "loss": 1.3166,
+      "step": 560
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.919439028925587e-05,
+      "loss": 1.3377,
+      "step": 561
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.919055733094774e-05,
+      "loss": 1.3287,
+      "step": 562
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.91867154642937e-05,
+      "loss": 1.3511,
+      "step": 563
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.918286469107718e-05,
+      "loss": 1.3573,
+      "step": 564
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.917900501308572e-05,
+      "loss": 1.3141,
+      "step": 565
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9175136432111e-05,
+      "loss": 1.3611,
+      "step": 566
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9171258949948827e-05,
+      "loss": 1.3282,
+      "step": 567
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.916737256839916e-05,
+      "loss": 1.3067,
+      "step": 568
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.916347728926606e-05,
+      "loss": 1.3499,
+      "step": 569
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.915957311435774e-05,
+      "loss": 1.2912,
+      "step": 570
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.915566004548654e-05,
+      "loss": 1.3769,
+      "step": 571
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.915173808446892e-05,
+      "loss": 1.2684,
+      "step": 572
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.914780723312548e-05,
+      "loss": 1.3661,
+      "step": 573
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.914386749328093e-05,
+      "loss": 1.3749,
+      "step": 574
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.913991886676412e-05,
+      "loss": 1.3324,
+      "step": 575
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9135961355408024e-05,
+      "loss": 1.2841,
+      "step": 576
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.913199496104972e-05,
+      "loss": 1.3472,
+      "step": 577
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.912801968553045e-05,
+      "loss": 1.317,
+      "step": 578
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9124035530695546e-05,
+      "loss": 1.3709,
+      "step": 579
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.912004249839447e-05,
+      "loss": 1.3045,
+      "step": 580
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.91160405904808e-05,
+      "loss": 1.3511,
+      "step": 581
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.911202980881226e-05,
+      "loss": 1.3596,
+      "step": 582
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.910801015525064e-05,
+      "loss": 1.3468,
+      "step": 583
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.910398163166192e-05,
+      "loss": 1.3168,
+      "step": 584
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.909994423991614e-05,
+      "loss": 1.3635,
+      "step": 585
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.909589798188747e-05,
+      "loss": 1.3195,
+      "step": 586
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.909184285945421e-05,
+      "loss": 1.3587,
+      "step": 587
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.908777887449877e-05,
+      "loss": 1.3586,
+      "step": 588
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9083706028907665e-05,
+      "loss": 1.3146,
+      "step": 589
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9079624324571536e-05,
+      "loss": 1.3445,
+      "step": 590
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9075533763385116e-05,
+      "loss": 1.4044,
+      "step": 591
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9071434347247275e-05,
+      "loss": 1.3535,
+      "step": 592
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.906732607806098e-05,
+      "loss": 1.2993,
+      "step": 593
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.906320895773329e-05,
+      "loss": 1.3229,
+      "step": 594
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.905908298817543e-05,
+      "loss": 1.353,
+      "step": 595
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.905494817130265e-05,
+      "loss": 1.3275,
+      "step": 596
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.9050804509034383e-05,
+      "loss": 1.3252,
+      "step": 597
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.904665200329411e-05,
+      "loss": 1.3534,
+      "step": 598
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.904249065600948e-05,
+      "loss": 1.365,
+      "step": 599
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.903832046911218e-05,
+      "loss": 1.3479,
+      "step": 600
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.9034141444538034e-05,
+      "loss": 1.3279,
+      "step": 601
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.902995358422697e-05,
+      "loss": 1.3512,
+      "step": 602
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.902575689012301e-05,
+      "loss": 1.332,
+      "step": 603
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.9021551364174286e-05,
+      "loss": 1.3107,
+      "step": 604
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.901733700833301e-05,
+      "loss": 1.2952,
+      "step": 605
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.9013113824555515e-05,
+      "loss": 1.2863,
+      "step": 606
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.9008881814802225e-05,
+      "loss": 1.3365,
+      "step": 607
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.900464098103765e-05,
+      "loss": 1.348,
+      "step": 608
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.9000391325230405e-05,
+      "loss": 1.3923,
+      "step": 609
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.899613284935321e-05,
+      "loss": 1.3236,
+      "step": 610
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.899186555538286e-05,
+      "loss": 1.346,
+      "step": 611
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.898758944530025e-05,
+      "loss": 1.3396,
+      "step": 612
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.898330452109038e-05,
+      "loss": 1.3161,
+      "step": 613
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.897901078474233e-05,
+      "loss": 1.3269,
+      "step": 614
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.897470823824927e-05,
+      "loss": 1.3353,
+      "step": 615
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.897039688360845e-05,
+      "loss": 1.3457,
+      "step": 616
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.8966076722821245e-05,
+      "loss": 1.393,
+      "step": 617
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.8961747757893075e-05,
+      "loss": 1.3907,
+      "step": 618
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.895740999083347e-05,
+      "loss": 1.3242,
+      "step": 619
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.8953063423656055e-05,
+      "loss": 1.2941,
+      "step": 620
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.8948708058378504e-05,
+      "loss": 1.3333,
+      "step": 621
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.894434389702261e-05,
+      "loss": 1.2923,
+      "step": 622
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.8939970941614247e-05,
+      "loss": 1.3301,
+      "step": 623
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.893558919418334e-05,
+      "loss": 1.332,
+      "step": 624
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.893119865676393e-05,
+      "loss": 1.3211,
+      "step": 625
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.892679933139412e-05,
+      "loss": 1.3203,
+      "step": 626
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.8922391220116094e-05,
+      "loss": 1.3262,
+      "step": 627
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.891797432497613e-05,
+      "loss": 1.325,
+      "step": 628
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.891354864802455e-05,
+      "loss": 1.3373,
+      "step": 629
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.89091141913158e-05,
+      "loss": 1.3773,
+      "step": 630
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.890467095690837e-05,
+      "loss": 1.3327,
+      "step": 631
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.890021894686481e-05,
+      "loss": 1.32,
+      "step": 632
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.8895758163251783e-05,
+      "loss": 1.3501,
+      "step": 633
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.889128860814e-05,
+      "loss": 1.3465,
+      "step": 634
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.8886810283604245e-05,
+      "loss": 1.3617,
+      "step": 635
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.888232319172338e-05,
+      "loss": 1.3613,
+      "step": 636
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.887782733458034e-05,
+      "loss": 1.313,
+      "step": 637
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.887332271426211e-05,
+      "loss": 1.3819,
+      "step": 638
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.886880933285977e-05,
+      "loss": 1.3359,
+      "step": 639
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.886428719246845e-05,
+      "loss": 1.2789,
+      "step": 640
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.885975629518734e-05,
+      "loss": 1.3289,
+      "step": 641
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8855216643119697e-05,
+      "loss": 1.389,
+      "step": 642
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.885066823837287e-05,
+      "loss": 1.3445,
+      "step": 643
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.884611108305824e-05,
+      "loss": 1.2976,
+      "step": 644
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8841545179291254e-05,
+      "loss": 1.3225,
+      "step": 645
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.883697052919143e-05,
+      "loss": 1.3322,
+      "step": 646
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.883238713488235e-05,
+      "loss": 1.3157,
+      "step": 647
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.882779499849163e-05,
+      "loss": 1.3124,
+      "step": 648
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8823194122150975e-05,
+      "loss": 1.3327,
+      "step": 649
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.881858450799612e-05,
+      "loss": 1.3617,
+      "step": 650
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8813966158166894e-05,
+      "loss": 1.3349,
+      "step": 651
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8809339074807125e-05,
+      "loss": 1.3099,
+      "step": 652
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8804703260064756e-05,
+      "loss": 1.3449,
+      "step": 653
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.880005871609173e-05,
+      "loss": 1.3623,
+      "step": 654
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.879540544504408e-05,
+      "loss": 1.3599,
+      "step": 655
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.879074344908187e-05,
+      "loss": 1.3152,
+      "step": 656
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.878607273036922e-05,
+      "loss": 1.3683,
+      "step": 657
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8781393291074296e-05,
+      "loss": 1.3504,
+      "step": 658
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8776705133369333e-05,
+      "loss": 1.3143,
+      "step": 659
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8772008259430575e-05,
+      "loss": 1.3358,
+      "step": 660
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.876730267143834e-05,
+      "loss": 1.3297,
+      "step": 661
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.876258837157699e-05,
+      "loss": 1.3036,
+      "step": 662
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8757865362034914e-05,
+      "loss": 1.3259,
+      "step": 663
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.875313364500456e-05,
+      "loss": 1.2908,
+      "step": 664
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8748393222682425e-05,
+      "loss": 1.3005,
+      "step": 665
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.874364409726901e-05,
+      "loss": 1.3045,
+      "step": 666
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.87388862709689e-05,
+      "loss": 1.3331,
+      "step": 667
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8734119745990696e-05,
+      "loss": 1.3348,
+      "step": 668
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.872934452454704e-05,
+      "loss": 1.3533,
+      "step": 669
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.872456060885461e-05,
+      "loss": 1.288,
+      "step": 670
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8719768001134124e-05,
+      "loss": 1.3604,
+      "step": 671
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.871496670361033e-05,
+      "loss": 1.3343,
+      "step": 672
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.871015671851202e-05,
+      "loss": 1.3566,
+      "step": 673
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.870533804807201e-05,
+      "loss": 1.354,
+      "step": 674
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.870051069452714e-05,
+      "loss": 1.3175,
+      "step": 675
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8695674660118294e-05,
+      "loss": 1.3104,
+      "step": 676
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8690829947090386e-05,
+      "loss": 1.3424,
+      "step": 677
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.868597655769235e-05,
+      "loss": 1.3718,
+      "step": 678
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.868111449417716e-05,
+      "loss": 1.3294,
+      "step": 679
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.867624375880179e-05,
+      "loss": 1.2584,
+      "step": 680
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8671364353827284e-05,
+      "loss": 1.369,
+      "step": 681
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8666476281518665e-05,
+      "loss": 1.3192,
+      "step": 682
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8661579544145e-05,
+      "loss": 1.3639,
+      "step": 683
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8656674143979386e-05,
+      "loss": 1.2207,
+      "step": 684
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8651760083298926e-05,
+      "loss": 1.3189,
+      "step": 685
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.864683736438475e-05,
+      "loss": 1.3861,
+      "step": 686
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8641905989522016e-05,
+      "loss": 1.3206,
+      "step": 687
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.863696596099988e-05,
+      "loss": 1.317,
+      "step": 688
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.863201728111153e-05,
+      "loss": 1.3675,
+      "step": 689
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.862705995215417e-05,
+      "loss": 1.3608,
+      "step": 690
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.862209397642901e-05,
+      "loss": 1.3174,
+      "step": 691
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.861711935624129e-05,
+      "loss": 1.3038,
+      "step": 692
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8612136093900224e-05,
+      "loss": 1.2865,
+      "step": 693
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.860714419171909e-05,
+      "loss": 1.3025,
+      "step": 694
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.860214365201515e-05,
+      "loss": 1.3254,
+      "step": 695
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8597134477109674e-05,
+      "loss": 1.2729,
+      "step": 696
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8592116669327945e-05,
+      "loss": 1.3217,
+      "step": 697
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.858709023099925e-05,
+      "loss": 1.3261,
+      "step": 698
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.858205516445689e-05,
+      "loss": 1.3223,
+      "step": 699
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.857701147203816e-05,
+      "loss": 1.3326,
+      "step": 700
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.857195915608437e-05,
+      "loss": 1.3546,
+      "step": 701
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8566898218940825e-05,
+      "loss": 1.344,
+      "step": 702
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.856182866295684e-05,
+      "loss": 1.3246,
+      "step": 703
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8556750490485724e-05,
+      "loss": 1.3234,
+      "step": 704
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.855166370388479e-05,
+      "loss": 1.3433,
+      "step": 705
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8546568305515345e-05,
+      "loss": 1.3348,
+      "step": 706
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.85414642977427e-05,
+      "loss": 1.3573,
+      "step": 707
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8536351682936155e-05,
+      "loss": 1.3409,
+      "step": 708
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8531230463469015e-05,
+      "loss": 1.3084,
+      "step": 709
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.852610064171857e-05,
+      "loss": 1.3788,
+      "step": 710
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.85209622200661e-05,
+      "loss": 1.3381,
+      "step": 711
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8515815200896905e-05,
+      "loss": 1.3015,
+      "step": 712
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.851065958660023e-05,
+      "loss": 1.3084,
+      "step": 713
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8505495379569354e-05,
+      "loss": 1.3429,
+      "step": 714
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.850032258220152e-05,
+      "loss": 1.3328,
+      "step": 715
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.849514119689796e-05,
+      "loss": 1.3462,
+      "step": 716
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.84899512260639e-05,
+      "loss": 1.2932,
+      "step": 717
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.848475267210856e-05,
+      "loss": 1.3455,
+      "step": 718
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8479545537445115e-05,
+      "loss": 1.348,
+      "step": 719
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.847432982449075e-05,
+      "loss": 1.3449,
+      "step": 720
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.846910553566662e-05,
+      "loss": 1.2975,
+      "step": 721
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.846387267339787e-05,
+      "loss": 1.3452,
+      "step": 722
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.845863124011361e-05,
+      "loss": 1.3705,
+      "step": 723
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.845338123824694e-05,
+      "loss": 1.3662,
+      "step": 724
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.844812267023495e-05,
+      "loss": 1.3419,
+      "step": 725
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8442855538518667e-05,
+      "loss": 1.3088,
+      "step": 726
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8437579845543133e-05,
+      "loss": 1.2827,
+      "step": 727
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.843229559375735e-05,
+      "loss": 1.2962,
+      "step": 728
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.842700278561429e-05,
+      "loss": 1.2687,
+      "step": 729
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8421701423570895e-05,
+      "loss": 1.2588,
+      "step": 730
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.841639151008809e-05,
+      "loss": 1.2686,
+      "step": 731
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8411073047630745e-05,
+      "loss": 1.3598,
+      "step": 732
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.840574603866774e-05,
+      "loss": 1.309,
+      "step": 733
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.840041048567188e-05,
+      "loss": 1.3183,
+      "step": 734
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.839506639111996e-05,
+      "loss": 1.3085,
+      "step": 735
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.838971375749272e-05,
+      "loss": 1.315,
+      "step": 736
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.83843525872749e-05,
+      "loss": 1.3152,
+      "step": 737
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.837898288295516e-05,
+      "loss": 1.2853,
+      "step": 738
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.837360464702616e-05,
+      "loss": 1.3565,
+      "step": 739
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8368217881984484e-05,
+      "loss": 1.2939,
+      "step": 740
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.83628225903307e-05,
+      "loss": 1.3342,
+      "step": 741
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8357418774569335e-05,
+      "loss": 1.2597,
+      "step": 742
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.835200643720886e-05,
+      "loss": 1.2928,
+      "step": 743
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8346585580761705e-05,
+      "loss": 1.3252,
+      "step": 744
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8341156207744254e-05,
+      "loss": 1.3273,
+      "step": 745
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.833571832067685e-05,
+      "loss": 1.3518,
+      "step": 746
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8330271922083795e-05,
+      "loss": 1.2859,
+      "step": 747
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8324817014493326e-05,
+      "loss": 1.3529,
+      "step": 748
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.831935360043763e-05,
+      "loss": 1.3298,
+      "step": 749
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8313881682452854e-05,
+      "loss": 1.2983,
+      "step": 750
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.830840126307909e-05,
+      "loss": 1.3388,
+      "step": 751
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.830291234486037e-05,
+      "loss": 1.355,
+      "step": 752
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8297414930344684e-05,
+      "loss": 1.3197,
+      "step": 753
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.829190902208394e-05,
+      "loss": 1.3258,
+      "step": 754
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.828639462263403e-05,
+      "loss": 1.3177,
+      "step": 755
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8280871734554746e-05,
+      "loss": 1.3312,
+      "step": 756
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.827534036040984e-05,
+      "loss": 1.3159,
+      "step": 757
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8269800502767e-05,
+      "loss": 1.3111,
+      "step": 758
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8264252164197866e-05,
+      "loss": 1.2511,
+      "step": 759
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.825869534727799e-05,
+      "loss": 1.3241,
+      "step": 760
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8253130054586886e-05,
+      "loss": 1.2966,
+      "step": 761
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.824755628870797e-05,
+      "loss": 1.3275,
+      "step": 762
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.824197405222863e-05,
+      "loss": 1.3324,
+      "step": 763
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8236383347740146e-05,
+      "loss": 1.3197,
+      "step": 764
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.823078417783777e-05,
+      "loss": 1.3572,
+      "step": 765
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8225176545120646e-05,
+      "loss": 1.3027,
+      "step": 766
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.821956045219186e-05,
+      "loss": 1.3135,
+      "step": 767
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.821393590165845e-05,
+      "loss": 1.3586,
+      "step": 768
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8208302896131344e-05,
+      "loss": 1.2514,
+      "step": 769
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.820266143822541e-05,
+      "loss": 1.3219,
+      "step": 770
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.819701153055944e-05,
+      "loss": 1.3377,
+      "step": 771
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8191353175756145e-05,
+      "loss": 1.3107,
+      "step": 772
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.818568637644217e-05,
+      "loss": 1.36,
+      "step": 773
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8180011135248055e-05,
+      "loss": 1.3692,
+      "step": 774
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8174327454808275e-05,
+      "loss": 1.3231,
+      "step": 775
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.816863533776124e-05,
+      "loss": 1.3213,
+      "step": 776
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.816293478674923e-05,
+      "loss": 1.3449,
+      "step": 777
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.815722580441849e-05,
+      "loss": 1.3097,
+      "step": 778
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.815150839341915e-05,
+      "loss": 1.3468,
+      "step": 779
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8145782556405244e-05,
+      "loss": 1.3213,
+      "step": 780
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.814004829603475e-05,
+      "loss": 1.3149,
+      "step": 781
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.813430561496953e-05,
+      "loss": 1.3353,
+      "step": 782
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.812855451587537e-05,
+      "loss": 1.3114,
+      "step": 783
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.812279500142194e-05,
+      "loss": 1.3304,
+      "step": 784
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.811702707428285e-05,
+      "loss": 1.3388,
+      "step": 785
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.81112507371356e-05,
+      "loss": 1.2986,
+      "step": 786
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.810546599266158e-05,
+      "loss": 1.3354,
+      "step": 787
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.8099672843546106e-05,
+      "loss": 1.3486,
+      "step": 788
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.809387129247838e-05,
+      "loss": 1.3586,
+      "step": 789
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.808806134215151e-05,
+      "loss": 1.3622,
+      "step": 790
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.80822429952625e-05,
+      "loss": 1.3256,
+      "step": 791
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.8076416254512256e-05,
+      "loss": 1.3196,
+      "step": 792
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.807058112260558e-05,
+      "loss": 1.2956,
+      "step": 793
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.8064737602251155e-05,
+      "loss": 1.3438,
+      "step": 794
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.8058885696161595e-05,
+      "loss": 1.34,
+      "step": 795
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.805302540705335e-05,
+      "loss": 1.3227,
+      "step": 796
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.8047156737646825e-05,
+      "loss": 1.3718,
+      "step": 797
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.8041279690666254e-05,
+      "loss": 1.2667,
+      "step": 798
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.803539426883982e-05,
+      "loss": 1.302,
+      "step": 799
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.8029500474899544e-05,
+      "loss": 1.341,
+      "step": 800
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.802359831158135e-05,
+      "loss": 1.343,
+      "step": 801
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.801768778162506e-05,
+      "loss": 1.3368,
+      "step": 802
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.8011768887774365e-05,
+      "loss": 1.3268,
+      "step": 803
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.800584163277684e-05,
+      "loss": 1.3112,
+      "step": 804
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.7999906019383954e-05,
+      "loss": 1.3869,
+      "step": 805
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.799396205035104e-05,
+      "loss": 1.3264,
+      "step": 806
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.7988009728437304e-05,
+      "loss": 1.3573,
+      "step": 807
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.7982049056405866e-05,
+      "loss": 1.3474,
+      "step": 808
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.797608003702368e-05,
+      "loss": 1.3319,
+      "step": 809
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.79701026730616e-05,
+      "loss": 1.3339,
+      "step": 810
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.796411696729434e-05,
+      "loss": 1.272,
+      "step": 811
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.79581229225005e-05,
+      "loss": 1.3628,
+      "step": 812
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.795212054146254e-05,
+      "loss": 1.3212,
+      "step": 813
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.794610982696679e-05,
+      "loss": 1.3288,
+      "step": 814
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.7940090781803454e-05,
+      "loss": 1.2961,
+      "step": 815
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.7934063408766606e-05,
+      "loss": 1.3084,
+      "step": 816
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.792802771065417e-05,
+      "loss": 1.328,
+      "step": 817
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.792198369026796e-05,
+      "loss": 1.2938,
+      "step": 818
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.791593135041362e-05,
+      "loss": 1.3381,
+      "step": 819
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.790987069390069e-05,
+      "loss": 1.2808,
+      "step": 820
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.790380172354255e-05,
+      "loss": 1.2853,
+      "step": 821
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.789772444215644e-05,
+      "loss": 1.3335,
+      "step": 822
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.7891638852563455e-05,
+      "loss": 1.3278,
+      "step": 823
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.788554495758858e-05,
+      "loss": 1.2816,
+      "step": 824
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.7879442760060604e-05,
+      "loss": 1.3327,
+      "step": 825
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.78733322628122e-05,
+      "loss": 1.3372,
+      "step": 826
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.786721346867991e-05,
+      "loss": 1.347,
+      "step": 827
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.786108638050408e-05,
+      "loss": 1.3139,
+      "step": 828
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.785495100112894e-05,
+      "loss": 1.3048,
+      "step": 829
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.784880733340257e-05,
+      "loss": 1.3011,
+      "step": 830
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.784265538017689e-05,
+      "loss": 1.3368,
+      "step": 831
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7836495144307644e-05,
+      "loss": 1.2558,
+      "step": 832
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.783032662865447e-05,
+      "loss": 1.3549,
+      "step": 833
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.782414983608081e-05,
+      "loss": 1.3048,
+      "step": 834
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7817964769453956e-05,
+      "loss": 1.326,
+      "step": 835
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.781177143164505e-05,
+      "loss": 1.3329,
+      "step": 836
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7805569825529055e-05,
+      "loss": 1.32,
+      "step": 837
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.779935995398481e-05,
+      "loss": 1.3174,
+      "step": 838
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7793141819894955e-05,
+      "loss": 1.3625,
+      "step": 839
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.778691542614596e-05,
+      "loss": 1.3777,
+      "step": 840
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.778068077562817e-05,
+      "loss": 1.3298,
+      "step": 841
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7774437871235724e-05,
+      "loss": 1.3167,
+      "step": 842
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.776818671586662e-05,
+      "loss": 1.2919,
+      "step": 843
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.776192731242265e-05,
+      "loss": 1.3498,
+      "step": 844
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.775565966380949e-05,
+      "loss": 1.3421,
+      "step": 845
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.774938377293659e-05,
+      "loss": 1.3182,
+      "step": 846
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.774309964271725e-05,
+      "loss": 1.3455,
+      "step": 847
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7736807276068604e-05,
+      "loss": 1.325,
+      "step": 848
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.773050667591158e-05,
+      "loss": 1.338,
+      "step": 849
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.772419784517095e-05,
+      "loss": 1.31,
+      "step": 850
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.771788078677532e-05,
+      "loss": 1.3155,
+      "step": 851
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.771155550365708e-05,
+      "loss": 1.2947,
+      "step": 852
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.770522199875247e-05,
+      "loss": 1.3655,
+      "step": 853
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7698880275001516e-05,
+      "loss": 1.3295,
+      "step": 854
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.769253033534808e-05,
+      "loss": 1.3509,
+      "step": 855
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7686172182739845e-05,
+      "loss": 1.3094,
+      "step": 856
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.767980582012828e-05,
+      "loss": 1.2749,
+      "step": 857
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7673431250468695e-05,
+      "loss": 1.3423,
+      "step": 858
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.766704847672018e-05,
+      "loss": 1.3708,
+      "step": 859
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.766065750184566e-05,
+      "loss": 1.3212,
+      "step": 860
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7654258328811856e-05,
+      "loss": 1.3406,
+      "step": 861
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.764785096058927e-05,
+      "loss": 1.317,
+      "step": 862
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.764143540015227e-05,
+      "loss": 1.2996,
+      "step": 863
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.763501165047896e-05,
+      "loss": 1.3164,
+      "step": 864
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7628579714551285e-05,
+      "loss": 1.3208,
+      "step": 865
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7622139595354976e-05,
+      "loss": 1.3605,
+      "step": 866
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7615691295879574e-05,
+      "loss": 1.3146,
+      "step": 867
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.76092348191184e-05,
+      "loss": 1.3372,
+      "step": 868
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7602770168068586e-05,
+      "loss": 1.302,
+      "step": 869
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.759629734573105e-05,
+      "loss": 1.3072,
+      "step": 870
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.758981635511051e-05,
+      "loss": 1.3293,
+      "step": 871
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.758332719921547e-05,
+      "loss": 1.3123,
+      "step": 872
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.757682988105823e-05,
+      "loss": 1.3352,
+      "step": 873
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7570324403654866e-05,
+      "loss": 1.2952,
+      "step": 874
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.756381077002526e-05,
+      "loss": 1.3695,
+      "step": 875
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.755728898319306e-05,
+      "loss": 1.3138,
+      "step": 876
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7550759046185726e-05,
+      "loss": 1.4168,
+      "step": 877
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7544220962034475e-05,
+      "loss": 1.3155,
+      "step": 878
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7537674733774315e-05,
+      "loss": 1.2633,
+      "step": 879
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.753112036444404e-05,
+      "loss": 1.3867,
+      "step": 880
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.752455785708622e-05,
+      "loss": 1.3262,
+      "step": 881
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7517987214747186e-05,
+      "loss": 1.3172,
+      "step": 882
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.751140844047708e-05,
+      "loss": 1.3271,
+      "step": 883
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7504821537329795e-05,
+      "loss": 1.304,
+      "step": 884
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7498226508362996e-05,
+      "loss": 1.3103,
+      "step": 885
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.749162335663813e-05,
+      "loss": 1.3819,
+      "step": 886
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7485012085220416e-05,
+      "loss": 1.3204,
+      "step": 887
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.747839269717882e-05,
+      "loss": 1.3505,
+      "step": 888
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7471765195586115e-05,
+      "loss": 1.3184,
+      "step": 889
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.74651295835188e-05,
+      "loss": 1.3156,
+      "step": 890
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.745848586405717e-05,
+      "loss": 1.311,
+      "step": 891
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.745183404028525e-05,
+      "loss": 1.306,
+      "step": 892
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7445174115290875e-05,
+      "loss": 1.3066,
+      "step": 893
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.74385060921656e-05,
+      "loss": 1.3126,
+      "step": 894
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.743182997400475e-05,
+      "loss": 1.2545,
+      "step": 895
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.742514576390741e-05,
+      "loss": 1.372,
+      "step": 896
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.741845346497643e-05,
+      "loss": 1.2692,
+      "step": 897
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.741175308031839e-05,
+      "loss": 1.2986,
+      "step": 898
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.740504461304366e-05,
+      "loss": 1.3442,
+      "step": 899
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.739832806626632e-05,
+      "loss": 1.2773,
+      "step": 900
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7391603443104244e-05,
+      "loss": 1.2644,
+      "step": 901
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.738487074667902e-05,
+      "loss": 1.3268,
+      "step": 902
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7378129980116e-05,
+      "loss": 1.3084,
+      "step": 903
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7371381146544276e-05,
+      "loss": 1.3093,
+      "step": 904
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.736462424909669e-05,
+      "loss": 1.2696,
+      "step": 905
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.735785929090983e-05,
+      "loss": 1.2589,
+      "step": 906
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7351086275124023e-05,
+      "loss": 1.2937,
+      "step": 907
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7344305204883326e-05,
+      "loss": 1.36,
+      "step": 908
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7337516083335536e-05,
+      "loss": 1.3503,
+      "step": 909
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7330718913632215e-05,
+      "loss": 1.3143,
+      "step": 910
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.732391369892862e-05,
+      "loss": 1.2691,
+      "step": 911
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.731710044238378e-05,
+      "loss": 1.3538,
+      "step": 912
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.731027914716044e-05,
+      "loss": 1.2826,
+      "step": 913
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7303449816425066e-05,
+      "loss": 1.3439,
+      "step": 914
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.729661245334787e-05,
+      "loss": 1.3079,
+      "step": 915
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.728976706110278e-05,
+      "loss": 1.2925,
+      "step": 916
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7282913642867484e-05,
+      "loss": 1.3079,
+      "step": 917
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.727605220182334e-05,
+      "loss": 1.3148,
+      "step": 918
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.726918274115548e-05,
+      "loss": 1.2858,
+      "step": 919
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.726230526405273e-05,
+      "loss": 1.332,
+      "step": 920
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.725541977370765e-05,
+      "loss": 1.3153,
+      "step": 921
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7248526273316524e-05,
+      "loss": 1.266,
+      "step": 922
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.724162476607933e-05,
+      "loss": 1.3352,
+      "step": 923
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.72347152551998e-05,
+      "loss": 1.2768,
+      "step": 924
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.722779774388535e-05,
+      "loss": 1.283,
+      "step": 925
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.722087223534711e-05,
+      "loss": 1.2801,
+      "step": 926
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.721393873279996e-05,
+      "loss": 1.3038,
+      "step": 927
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.720699723946244e-05,
+      "loss": 1.3247,
+      "step": 928
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.720004775855684e-05,
+      "loss": 1.2946,
+      "step": 929
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.719309029330912e-05,
+      "loss": 1.2982,
+      "step": 930
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7186124846948995e-05,
+      "loss": 1.2819,
+      "step": 931
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7179151422709845e-05,
+      "loss": 1.3325,
+      "step": 932
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.717217002382875e-05,
+      "loss": 1.3033,
+      "step": 933
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.716518065354654e-05,
+      "loss": 1.3273,
+      "step": 934
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.715818331510769e-05,
+      "loss": 1.314,
+      "step": 935
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.71511780117604e-05,
+      "loss": 1.3593,
+      "step": 936
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.714416474675657e-05,
+      "loss": 1.3345,
+      "step": 937
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7137143523351787e-05,
+      "loss": 1.3185,
+      "step": 938
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.713011434480534e-05,
+      "loss": 1.2981,
+      "step": 939
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.71230772143802e-05,
+      "loss": 1.2702,
+      "step": 940
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.711603213534303e-05,
+      "loss": 1.346,
+      "step": 941
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.710897911096421e-05,
+      "loss": 1.2751,
+      "step": 942
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.710191814451777e-05,
+      "loss": 1.307,
+      "step": 943
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7094849239281444e-05,
+      "loss": 1.3397,
+      "step": 944
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7087772398536656e-05,
+      "loss": 1.3408,
+      "step": 945
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.70806876255685e-05,
+      "loss": 1.2815,
+      "step": 946
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7073594923665774e-05,
+      "loss": 1.3294,
+      "step": 947
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7066494296120935e-05,
+      "loss": 1.3056,
+      "step": 948
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.705938574623012e-05,
+      "loss": 1.3195,
+      "step": 949
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.705226927729317e-05,
+      "loss": 1.3343,
+      "step": 950
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.704514489261357e-05,
+      "loss": 1.2853,
+      "step": 951
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.703801259549848e-05,
+      "loss": 1.3278,
+      "step": 952
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7030872389258777e-05,
+      "loss": 1.2837,
+      "step": 953
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.702372427720895e-05,
+      "loss": 1.3047,
+      "step": 954
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.701656826266721e-05,
+      "loss": 1.3364,
+      "step": 955
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7009404348955385e-05,
+      "loss": 1.3342,
+      "step": 956
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7002232539399014e-05,
+      "loss": 1.3222,
+      "step": 957
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.6995052837327274e-05,
+      "loss": 1.335,
+      "step": 958
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.6987865246073035e-05,
+      "loss": 1.3284,
+      "step": 959
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.6980669768972795e-05,
+      "loss": 1.3216,
+      "step": 960
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.6973466409366735e-05,
+      "loss": 1.3034,
+      "step": 961
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.696625517059868e-05,
+      "loss": 1.3074,
+      "step": 962
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.695903605601612e-05,
+      "loss": 1.3484,
+      "step": 963
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.695180906897021e-05,
+      "loss": 1.3161,
+      "step": 964
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.694457421281575e-05,
+      "loss": 1.3406,
+      "step": 965
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.693733149091119e-05,
+      "loss": 1.3661,
+      "step": 966
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.693008090661864e-05,
+      "loss": 1.2985,
+      "step": 967
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.6922822463303846e-05,
+      "loss": 1.3351,
+      "step": 968
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.691555616433622e-05,
+      "loss": 1.3145,
+      "step": 969
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.69082820130888e-05,
+      "loss": 1.3249,
+      "step": 970
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.69010000129383e-05,
+      "loss": 1.2538,
+      "step": 971
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.689371016726504e-05,
+      "loss": 1.3523,
+      "step": 972
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.6886412479453004e-05,
+      "loss": 1.3724,
+      "step": 973
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.6879106952889826e-05,
+      "loss": 1.3468,
+      "step": 974
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.687179359096675e-05,
+      "loss": 1.3211,
+      "step": 975
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.686447239707868e-05,
+      "loss": 1.3299,
+      "step": 976
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.685714337462415e-05,
+      "loss": 1.2972,
+      "step": 977
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6849806527005316e-05,
+      "loss": 1.2821,
+      "step": 978
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6842461857627986e-05,
+      "loss": 1.3079,
+      "step": 979
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6835109369901586e-05,
+      "loss": 1.2907,
+      "step": 980
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.682774906723918e-05,
+      "loss": 1.2979,
+      "step": 981
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6820380953057446e-05,
+      "loss": 1.331,
+      "step": 982
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.681300503077671e-05,
+      "loss": 1.2952,
+      "step": 983
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.680562130382089e-05,
+      "loss": 1.2752,
+      "step": 984
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.679822977561756e-05,
+      "loss": 1.3218,
+      "step": 985
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.67908304495979e-05,
+      "loss": 1.3119,
+      "step": 986
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.678342332919671e-05,
+      "loss": 1.281,
+      "step": 987
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6776008417852415e-05,
+      "loss": 1.2747,
+      "step": 988
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.676858571900704e-05,
+      "loss": 1.3361,
+      "step": 989
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6761155236106246e-05,
+      "loss": 1.3624,
+      "step": 990
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.67537169725993e-05,
+      "loss": 1.2927,
+      "step": 991
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6746270931939064e-05,
+      "loss": 1.3159,
+      "step": 992
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6738817117582045e-05,
+      "loss": 1.2895,
+      "step": 993
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6731355532988315e-05,
+      "loss": 1.3099,
+      "step": 994
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6723886181621595e-05,
+      "loss": 1.3087,
+      "step": 995
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6716409066949184e-05,
+      "loss": 1.2716,
+      "step": 996
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.670892419244199e-05,
+      "loss": 1.2939,
+      "step": 997
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.670143156157454e-05,
+      "loss": 1.3464,
+      "step": 998
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6693931177824934e-05,
+      "loss": 1.2778,
+      "step": 999
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.66864230446749e-05,
+      "loss": 1.3137,
+      "step": 1000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.667890716560973e-05,
+      "loss": 1.308,
+      "step": 1001
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.667138354411834e-05,
+      "loss": 1.3033,
+      "step": 1002
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.666385218369324e-05,
+      "loss": 1.3094,
+      "step": 1003
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6656313087830505e-05,
+      "loss": 1.3138,
+      "step": 1004
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.664876626002982e-05,
+      "loss": 1.3232,
+      "step": 1005
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6641211703794466e-05,
+      "loss": 1.331,
+      "step": 1006
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.66336494226313e-05,
+      "loss": 1.322,
+      "step": 1007
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.662607942005077e-05,
+      "loss": 1.3342,
+      "step": 1008
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.66185016995669e-05,
+      "loss": 1.3037,
+      "step": 1009
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.661091626469731e-05,
+      "loss": 1.3338,
+      "step": 1010
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6603323118963194e-05,
+      "loss": 1.2913,
+      "step": 1011
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.659572226588932e-05,
+      "loss": 1.3209,
+      "step": 1012
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.658811370900404e-05,
+      "loss": 1.2759,
+      "step": 1013
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.658049745183928e-05,
+      "loss": 1.3323,
+      "step": 1014
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.657287349793056e-05,
+      "loss": 1.3535,
+      "step": 1015
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.656524185081693e-05,
+      "loss": 1.2524,
+      "step": 1016
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.655760251404105e-05,
+      "loss": 1.3218,
+      "step": 1017
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.654995549114913e-05,
+      "loss": 1.3505,
+      "step": 1018
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6542300785690954e-05,
+      "loss": 1.2841,
+      "step": 1019
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6534638401219874e-05,
+      "loss": 1.3382,
+      "step": 1020
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.652696834129281e-05,
+      "loss": 1.3212,
+      "step": 1021
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6519290609470225e-05,
+      "loss": 1.3338,
+      "step": 1022
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.651160520931617e-05,
+      "loss": 1.2861,
+      "step": 1023
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.650391214439825e-05,
+      "loss": 1.2802,
+      "step": 1024
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.64962114182876e-05,
+      "loss": 1.2893,
+      "step": 1025
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.648850303455895e-05,
+      "loss": 1.3217,
+      "step": 1026
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6480786996790554e-05,
+      "loss": 1.3497,
+      "step": 1027
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.647306330856425e-05,
+      "loss": 1.3243,
+      "step": 1028
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.646533197346539e-05,
+      "loss": 1.298,
+      "step": 1029
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6457592995082915e-05,
+      "loss": 1.2769,
+      "step": 1030
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.644984637700928e-05,
+      "loss": 1.33,
+      "step": 1031
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6442092122840505e-05,
+      "loss": 1.327,
+      "step": 1032
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.643433023617616e-05,
+      "loss": 1.2911,
+      "step": 1033
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.642656072061933e-05,
+      "loss": 1.2925,
+      "step": 1034
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.641878357977668e-05,
+      "loss": 1.3223,
+      "step": 1035
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.641099881725839e-05,
+      "loss": 1.3262,
+      "step": 1036
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6403206436678173e-05,
+      "loss": 1.3007,
+      "step": 1037
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.63954064416533e-05,
+      "loss": 1.3166,
+      "step": 1038
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6387598835804555e-05,
+      "loss": 1.3546,
+      "step": 1039
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6379783622756275e-05,
+      "loss": 1.3309,
+      "step": 1040
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6371960806136313e-05,
+      "loss": 1.341,
+      "step": 1041
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.636413038957605e-05,
+      "loss": 1.319,
+      "step": 1042
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.635629237671041e-05,
+      "loss": 1.2956,
+      "step": 1043
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.634844677117784e-05,
+      "loss": 1.3025,
+      "step": 1044
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.63405935766203e-05,
+      "loss": 1.3476,
+      "step": 1045
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.633273279668327e-05,
+      "loss": 1.3185,
+      "step": 1046
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.632486443501578e-05,
+      "loss": 1.311,
+      "step": 1047
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.631698849527034e-05,
+      "loss": 1.3141,
+      "step": 1048
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.630910498110302e-05,
+      "loss": 1.3186,
+      "step": 1049
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.630121389617336e-05,
+      "loss": 1.3305,
+      "step": 1050
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.629331524414446e-05,
+      "loss": 1.3101,
+      "step": 1051
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6285409028682895e-05,
+      "loss": 1.2645,
+      "step": 1052
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.627749525345878e-05,
+      "loss": 1.3466,
+      "step": 1053
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.626957392214571e-05,
+      "loss": 1.3205,
+      "step": 1054
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.626164503842082e-05,
+      "loss": 1.3256,
+      "step": 1055
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6253708605964724e-05,
+      "loss": 1.2844,
+      "step": 1056
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6245764628461556e-05,
+      "loss": 1.3179,
+      "step": 1057
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6237813109598944e-05,
+      "loss": 1.3443,
+      "step": 1058
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.622985405306803e-05,
+      "loss": 1.2955,
+      "step": 1059
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.622188746256343e-05,
+      "loss": 1.3075,
+      "step": 1060
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.621391334178328e-05,
+      "loss": 1.3125,
+      "step": 1061
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.62059316944292e-05,
+      "loss": 1.2941,
+      "step": 1062
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.619794252420632e-05,
+      "loss": 1.2985,
+      "step": 1063
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.618994583482323e-05,
+      "loss": 1.3151,
+      "step": 1064
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.618194162999205e-05,
+      "loss": 1.296,
+      "step": 1065
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.617392991342836e-05,
+      "loss": 1.3287,
+      "step": 1066
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.616591068885123e-05,
+      "loss": 1.2997,
+      "step": 1067
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6157883959983234e-05,
+      "loss": 1.2501,
+      "step": 1068
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.614984973055041e-05,
+      "loss": 1.3166,
+      "step": 1069
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.614180800428228e-05,
+      "loss": 1.3497,
+      "step": 1070
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.6133758784911864e-05,
+      "loss": 1.3219,
+      "step": 1071
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.6125702076175636e-05,
+      "loss": 1.3273,
+      "step": 1072
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.611763788181356e-05,
+      "loss": 1.2938,
+      "step": 1073
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.610956620556907e-05,
+      "loss": 1.3263,
+      "step": 1074
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.610148705118908e-05,
+      "loss": 1.3174,
+      "step": 1075
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.609340042242397e-05,
+      "loss": 1.2726,
+      "step": 1076
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.6085306323027596e-05,
+      "loss": 1.3284,
+      "step": 1077
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.607720475675727e-05,
+      "loss": 1.3459,
+      "step": 1078
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.606909572737378e-05,
+      "loss": 1.3363,
+      "step": 1079
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.6060979238641363e-05,
+      "loss": 1.2867,
+      "step": 1080
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.6052855294327746e-05,
+      "loss": 1.3195,
+      "step": 1081
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.604472389820409e-05,
+      "loss": 1.2758,
+      "step": 1082
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.6036585054045044e-05,
+      "loss": 1.3045,
+      "step": 1083
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.602843876562868e-05,
+      "loss": 1.3325,
+      "step": 1084
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.6020285036736554e-05,
+      "loss": 1.2989,
+      "step": 1085
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.601212387115366e-05,
+      "loss": 1.3362,
+      "step": 1086
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.6003955272668444e-05,
+      "loss": 1.314,
+      "step": 1087
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5995779245072816e-05,
+      "loss": 1.321,
+      "step": 1088
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5987595792162126e-05,
+      "loss": 1.3526,
+      "step": 1089
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.597940491773516e-05,
+      "loss": 1.2579,
+      "step": 1090
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5971206625594176e-05,
+      "loss": 1.333,
+      "step": 1091
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5963000919544844e-05,
+      "loss": 1.3384,
+      "step": 1092
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.59547878033963e-05,
+      "loss": 1.2867,
+      "step": 1093
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.594656728096111e-05,
+      "loss": 1.3241,
+      "step": 1094
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5938339356055274e-05,
+      "loss": 1.3063,
+      "step": 1095
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.593010403249824e-05,
+      "loss": 1.316,
+      "step": 1096
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.592186131411288e-05,
+      "loss": 1.3018,
+      "step": 1097
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5913611204725496e-05,
+      "loss": 1.3229,
+      "step": 1098
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.590535370816584e-05,
+      "loss": 1.3447,
+      "step": 1099
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.589708882826707e-05,
+      "loss": 1.3017,
+      "step": 1100
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.588881656886578e-05,
+      "loss": 1.2863,
+      "step": 1101
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5880536933802e-05,
+      "loss": 1.3163,
+      "step": 1102
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.587224992691917e-05,
+      "loss": 1.3418,
+      "step": 1103
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.586395555206417e-05,
+      "loss": 1.3082,
+      "step": 1104
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.585565381308726e-05,
+      "loss": 1.2337,
+      "step": 1105
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.584734471384217e-05,
+      "loss": 1.2875,
+      "step": 1106
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5839028258186014e-05,
+      "loss": 1.3158,
+      "step": 1107
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.583070444997932e-05,
+      "loss": 1.3193,
+      "step": 1108
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5822373293086055e-05,
+      "loss": 1.3294,
+      "step": 1109
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.581403479137358e-05,
+      "loss": 1.2849,
+      "step": 1110
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.580568894871265e-05,
+      "loss": 1.3061,
+      "step": 1111
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.579733576897746e-05,
+      "loss": 1.2782,
+      "step": 1112
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.578897525604558e-05,
+      "loss": 1.3421,
+      "step": 1113
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.578060741379801e-05,
+      "loss": 1.3636,
+      "step": 1114
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.577223224611915e-05,
+      "loss": 1.2847,
+      "step": 1115
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.576384975689677e-05,
+      "loss": 1.2856,
+      "step": 1116
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.575545995002207e-05,
+      "loss": 1.3305,
+      "step": 1117
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.574706282938964e-05,
+      "loss": 1.3576,
+      "step": 1118
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.573865839889746e-05,
+      "loss": 1.3519,
+      "step": 1119
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5730246662446916e-05,
+      "loss": 1.2934,
+      "step": 1120
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.572182762394276e-05,
+      "loss": 1.3416,
+      "step": 1121
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.571340128729315e-05,
+      "loss": 1.2764,
+      "step": 1122
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.570496765640964e-05,
+      "loss": 1.2825,
+      "step": 1123
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.569652673520715e-05,
+      "loss": 1.3168,
+      "step": 1124
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5688078527604e-05,
+      "loss": 1.259,
+      "step": 1125
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.56796230375219e-05,
+      "loss": 1.2608,
+      "step": 1126
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.567116026888591e-05,
+      "loss": 1.3269,
+      "step": 1127
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5662690225624484e-05,
+      "loss": 1.2956,
+      "step": 1128
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.565421291166946e-05,
+      "loss": 1.2896,
+      "step": 1129
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5645728330956074e-05,
+      "loss": 1.3015,
+      "step": 1130
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.563723648742286e-05,
+      "loss": 1.3056,
+      "step": 1131
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5628737385011814e-05,
+      "loss": 1.2903,
+      "step": 1132
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.562023102766822e-05,
+      "loss": 1.3046,
+      "step": 1133
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.561171741934081e-05,
+      "loss": 1.3552,
+      "step": 1134
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.56031965639816e-05,
+      "loss": 1.3289,
+      "step": 1135
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.559466846554604e-05,
+      "loss": 1.3383,
+      "step": 1136
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5586133127992904e-05,
+      "loss": 1.3431,
+      "step": 1137
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.557759055528433e-05,
+      "loss": 1.2967,
+      "step": 1138
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5569040751385825e-05,
+      "loss": 1.3167,
+      "step": 1139
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.556048372026625e-05,
+      "loss": 1.2515,
+      "step": 1140
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.555191946589781e-05,
+      "loss": 1.3347,
+      "step": 1141
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.554334799225608e-05,
+      "loss": 1.3092,
+      "step": 1142
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.553476930331996e-05,
+      "loss": 1.3033,
+      "step": 1143
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5526183403071754e-05,
+      "loss": 1.2948,
+      "step": 1144
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.551759029549705e-05,
+      "loss": 1.3345,
+      "step": 1145
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.550898998458481e-05,
+      "loss": 1.2931,
+      "step": 1146
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.550038247432734e-05,
+      "loss": 1.3501,
+      "step": 1147
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.549176776872029e-05,
+      "loss": 1.3781,
+      "step": 1148
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5483145871762646e-05,
+      "loss": 1.3182,
+      "step": 1149
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.547451678745673e-05,
+      "loss": 1.3186,
+      "step": 1150
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.54658805198082e-05,
+      "loss": 1.3037,
+      "step": 1151
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.545723707282606e-05,
+      "loss": 1.2581,
+      "step": 1152
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5448586450522635e-05,
+      "loss": 1.3256,
+      "step": 1153
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5439928656913586e-05,
+      "loss": 1.3381,
+      "step": 1154
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.543126369601789e-05,
+      "loss": 1.2692,
+      "step": 1155
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.542259157185787e-05,
+      "loss": 1.3142,
+      "step": 1156
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5413912288459174e-05,
+      "loss": 1.3083,
+      "step": 1157
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5405225849850754e-05,
+      "loss": 1.3383,
+      "step": 1158
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.53965322600649e-05,
+      "loss": 1.2736,
+      "step": 1159
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5387831523137216e-05,
+      "loss": 1.3476,
+      "step": 1160
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5379123643106625e-05,
+      "loss": 1.2945,
+      "step": 1161
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5370408624015364e-05,
+      "loss": 1.3525,
+      "step": 1162
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.536168646990899e-05,
+      "loss": 1.3644,
+      "step": 1163
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.535295718483636e-05,
+      "loss": 1.3243,
+      "step": 1164
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5344220772849654e-05,
+      "loss": 1.3574,
+      "step": 1165
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.533547723800435e-05,
+      "loss": 1.3056,
+      "step": 1166
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.532672658435925e-05,
+      "loss": 1.2936,
+      "step": 1167
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.531796881597643e-05,
+      "loss": 1.3844,
+      "step": 1168
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.53092039369213e-05,
+      "loss": 1.3335,
+      "step": 1169
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.530043195126255e-05,
+      "loss": 1.2566,
+      "step": 1170
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.529165286307219e-05,
+      "loss": 1.3402,
+      "step": 1171
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.528286667642549e-05,
+      "loss": 1.314,
+      "step": 1172
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.527407339540106e-05,
+      "loss": 1.3547,
+      "step": 1173
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5265273024080776e-05,
+      "loss": 1.291,
+      "step": 1174
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.52564655665498e-05,
+      "loss": 1.3086,
+      "step": 1175
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.524765102689662e-05,
+      "loss": 1.3185,
+      "step": 1176
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.523882940921296e-05,
+      "loss": 1.3319,
+      "step": 1177
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5230000717593865e-05,
+      "loss": 1.299,
+      "step": 1178
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.522116495613766e-05,
+      "loss": 1.3019,
+      "step": 1179
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.521232212894594e-05,
+      "loss": 1.2854,
+      "step": 1180
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5203472240123594e-05,
+      "loss": 1.3305,
+      "step": 1181
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.519461529377877e-05,
+      "loss": 1.308,
+      "step": 1182
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5185751294022914e-05,
+      "loss": 1.2531,
+      "step": 1183
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5176880244970735e-05,
+      "loss": 1.2869,
+      "step": 1184
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.516800215074021e-05,
+      "loss": 1.3048,
+      "step": 1185
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.515911701545259e-05,
+      "loss": 1.3049,
+      "step": 1186
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5150224843232405e-05,
+      "loss": 1.2604,
+      "step": 1187
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.514132563820744e-05,
+      "loss": 1.3201,
+      "step": 1188
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.513241940450874e-05,
+      "loss": 1.3027,
+      "step": 1189
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.512350614627062e-05,
+      "loss": 1.2974,
+      "step": 1190
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.511458586763067e-05,
+      "loss": 1.2961,
+      "step": 1191
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.510565857272972e-05,
+      "loss": 1.2657,
+      "step": 1192
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.509672426571185e-05,
+      "loss": 1.2957,
+      "step": 1193
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.508778295072441e-05,
+      "loss": 1.3012,
+      "step": 1194
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5078834631918014e-05,
+      "loss": 1.3182,
+      "step": 1195
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.506987931344649e-05,
+      "loss": 1.3301,
+      "step": 1196
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.506091699946697e-05,
+      "loss": 1.2858,
+      "step": 1197
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.505194769413977e-05,
+      "loss": 1.3322,
+      "step": 1198
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.504297140162851e-05,
+      "loss": 1.2556,
+      "step": 1199
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.50339881261e-05,
+      "loss": 1.2901,
+      "step": 1200
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.502499787172434e-05,
+      "loss": 1.3396,
+      "step": 1201
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5016000642674836e-05,
+      "loss": 1.3205,
+      "step": 1202
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.500699644312805e-05,
+      "loss": 1.3128,
+      "step": 1203
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.4997985277263765e-05,
+      "loss": 1.304,
+      "step": 1204
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.498896714926502e-05,
+      "loss": 1.3031,
+      "step": 1205
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.4979942063318066e-05,
+      "loss": 1.333,
+      "step": 1206
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.497091002361238e-05,
+      "loss": 1.3031,
+      "step": 1207
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.496187103434069e-05,
+      "loss": 1.3285,
+      "step": 1208
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.4952825099698926e-05,
+      "loss": 1.3237,
+      "step": 1209
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.4943772223886264e-05,
+      "loss": 1.3202,
+      "step": 1210
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.493471241110507e-05,
+      "loss": 1.3213,
+      "step": 1211
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.492564566556098e-05,
+      "loss": 1.3025,
+      "step": 1212
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.491657199146281e-05,
+      "loss": 1.314,
+      "step": 1213
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.490749139302258e-05,
+      "loss": 1.3774,
+      "step": 1214
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4898403874455584e-05,
+      "loss": 1.3192,
+      "step": 1215
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4889309439980256e-05,
+      "loss": 1.2976,
+      "step": 1216
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.488020809381829e-05,
+      "loss": 1.334,
+      "step": 1217
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4871099840194575e-05,
+      "loss": 1.29,
+      "step": 1218
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4861984683337205e-05,
+      "loss": 1.2634,
+      "step": 1219
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.485286262747747e-05,
+      "loss": 1.2728,
+      "step": 1220
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4843733676849876e-05,
+      "loss": 1.3374,
+      "step": 1221
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4834597835692117e-05,
+      "loss": 1.2964,
+      "step": 1222
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.482545510824511e-05,
+      "loss": 1.3318,
+      "step": 1223
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.481630549875293e-05,
+      "loss": 1.2775,
+      "step": 1224
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.480714901146289e-05,
+      "loss": 1.335,
+      "step": 1225
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.479798565062546e-05,
+      "loss": 1.2692,
+      "step": 1226
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.478881542049432e-05,
+      "loss": 1.3023,
+      "step": 1227
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4779638325326326e-05,
+      "loss": 1.3615,
+      "step": 1228
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.477045436938154e-05,
+      "loss": 1.3302,
+      "step": 1229
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.476126355692318e-05,
+      "loss": 1.3058,
+      "step": 1230
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.475206589221768e-05,
+      "loss": 1.2731,
+      "step": 1231
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4742861379534636e-05,
+      "loss": 1.3252,
+      "step": 1232
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.473365002314682e-05,
+      "loss": 1.3084,
+      "step": 1233
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4724431827330196e-05,
+      "loss": 1.2977,
+      "step": 1234
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4715206796363876e-05,
+      "loss": 1.2764,
+      "step": 1235
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.470597493453018e-05,
+      "loss": 1.31,
+      "step": 1236
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.469673624611457e-05,
+      "loss": 1.2913,
+      "step": 1237
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4687490735405696e-05,
+      "loss": 1.2798,
+      "step": 1238
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.467823840669536e-05,
+      "loss": 1.346,
+      "step": 1239
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.466897926427854e-05,
+      "loss": 1.2654,
+      "step": 1240
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.465971331245337e-05,
+      "loss": 1.302,
+      "step": 1241
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.465044055552116e-05,
+      "loss": 1.2815,
+      "step": 1242
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.464116099778634e-05,
+      "loss": 1.2751,
+      "step": 1243
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.463187464355655e-05,
+      "loss": 1.3094,
+      "step": 1244
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.462258149714255e-05,
+      "loss": 1.3194,
+      "step": 1245
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.461328156285826e-05,
+      "loss": 1.2753,
+      "step": 1246
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4603974845020754e-05,
+      "loss": 1.3322,
+      "step": 1247
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4594661347950255e-05,
+      "loss": 1.372,
+      "step": 1248
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.458534107597013e-05,
+      "loss": 1.311,
+      "step": 1249
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.457601403340689e-05,
+      "loss": 1.3296,
+      "step": 1250
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.45666802245902e-05,
+      "loss": 1.317,
+      "step": 1251
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.455733965385284e-05,
+      "loss": 1.2959,
+      "step": 1252
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.454799232553077e-05,
+      "loss": 1.3007,
+      "step": 1253
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4538638243963045e-05,
+      "loss": 1.2638,
+      "step": 1254
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4529277413491885e-05,
+      "loss": 1.3071,
+      "step": 1255
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.451990983846262e-05,
+      "loss": 1.3229,
+      "step": 1256
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.451053552322373e-05,
+      "loss": 1.3249,
+      "step": 1257
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4501154472126815e-05,
+      "loss": 1.3593,
+      "step": 1258
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4491766689526596e-05,
+      "loss": 1.2989,
+      "step": 1259
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.448237217978093e-05,
+      "loss": 1.2947,
+      "step": 1260
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4472970947250794e-05,
+      "loss": 1.2887,
+      "step": 1261
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.446356299630028e-05,
+      "loss": 1.341,
+      "step": 1262
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4454148331296606e-05,
+      "loss": 1.3018,
+      "step": 1263
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.44447269566101e-05,
+      "loss": 1.3284,
+      "step": 1264
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.443529887661421e-05,
+      "loss": 1.3196,
+      "step": 1265
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.44258640956855e-05,
+      "loss": 1.3603,
+      "step": 1266
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.441642261820363e-05,
+      "loss": 1.2977,
+      "step": 1267
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.440697444855137e-05,
+      "loss": 1.3293,
+      "step": 1268
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.439751959111463e-05,
+      "loss": 1.3461,
+      "step": 1269
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.438805805028238e-05,
+      "loss": 1.3195,
+      "step": 1270
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4378589830446714e-05,
+      "loss": 1.2863,
+      "step": 1271
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.436911493600282e-05,
+      "loss": 1.2776,
+      "step": 1272
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4359633371349e-05,
+      "loss": 1.3062,
+      "step": 1273
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.435014514088662e-05,
+      "loss": 1.2985,
+      "step": 1274
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.434065024902019e-05,
+      "loss": 1.2872,
+      "step": 1275
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4331148700157263e-05,
+      "loss": 1.2782,
+      "step": 1276
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.43216404987085e-05,
+      "loss": 1.3008,
+      "step": 1277
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4312125649087664e-05,
+      "loss": 1.2953,
+      "step": 1278
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.430260415571158e-05,
+      "loss": 1.3191,
+      "step": 1279
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.429307602300019e-05,
+      "loss": 1.2763,
+      "step": 1280
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4283541255376466e-05,
+      "loss": 1.2801,
+      "step": 1281
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.427399985726652e-05,
+      "loss": 1.3344,
+      "step": 1282
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.42644518330995e-05,
+      "loss": 1.2885,
+      "step": 1283
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.425489718730765e-05,
+      "loss": 1.3324,
+      "step": 1284
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4245335924326274e-05,
+      "loss": 1.3043,
+      "step": 1285
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.423576804859375e-05,
+      "loss": 1.2856,
+      "step": 1286
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.422619356455154e-05,
+      "loss": 1.3455,
+      "step": 1287
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.421661247664417e-05,
+      "loss": 1.3051,
+      "step": 1288
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.420702478931921e-05,
+      "loss": 1.3006,
+      "step": 1289
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.419743050702732e-05,
+      "loss": 1.2823,
+      "step": 1290
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4187829634222205e-05,
+      "loss": 1.3047,
+      "step": 1291
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.417822217536064e-05,
+      "loss": 1.3242,
+      "step": 1292
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4168608134902443e-05,
+      "loss": 1.2822,
+      "step": 1293
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4158987517310506e-05,
+      "loss": 1.2905,
+      "step": 1294
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.414936032705075e-05,
+      "loss": 1.3127,
+      "step": 1295
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.413972656859218e-05,
+      "loss": 1.3153,
+      "step": 1296
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.413008624640683e-05,
+      "loss": 1.3078,
+      "step": 1297
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4120439364969766e-05,
+      "loss": 1.2792,
+      "step": 1298
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.411078592875912e-05,
+      "loss": 1.3039,
+      "step": 1299
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.410112594225607e-05,
+      "loss": 1.2918,
+      "step": 1300
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4091459409944836e-05,
+      "loss": 1.2701,
+      "step": 1301
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.408178633631265e-05,
+      "loss": 1.2595,
+      "step": 1302
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.407210672584979e-05,
+      "loss": 1.3252,
+      "step": 1303
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.40624205830496e-05,
+      "loss": 1.2767,
+      "step": 1304
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4052727912408414e-05,
+      "loss": 1.3054,
+      "step": 1305
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.404302871842563e-05,
+      "loss": 1.3169,
+      "step": 1306
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.403332300560364e-05,
+      "loss": 1.2939,
+      "step": 1307
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.4023610778447895e-05,
+      "loss": 1.3022,
+      "step": 1308
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.401389204146685e-05,
+      "loss": 1.3017,
+      "step": 1309
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.4004166799172004e-05,
+      "loss": 1.3318,
+      "step": 1310
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3994435056077827e-05,
+      "loss": 1.3287,
+      "step": 1311
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3984696816701865e-05,
+      "loss": 1.2773,
+      "step": 1312
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.397495208556465e-05,
+      "loss": 1.3318,
+      "step": 1313
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3965200867189734e-05,
+      "loss": 1.3425,
+      "step": 1314
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.395544316610367e-05,
+      "loss": 1.3127,
+      "step": 1315
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.394567898683602e-05,
+      "loss": 1.2811,
+      "step": 1316
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3935908333919385e-05,
+      "loss": 1.3013,
+      "step": 1317
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3926131211889336e-05,
+      "loss": 1.2883,
+      "step": 1318
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.391634762528445e-05,
+      "loss": 1.2957,
+      "step": 1319
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3906557578646317e-05,
+      "loss": 1.2667,
+      "step": 1320
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.389676107651953e-05,
+      "loss": 1.3282,
+      "step": 1321
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.388695812345168e-05,
+      "loss": 1.3189,
+      "step": 1322
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3877148723993306e-05,
+      "loss": 1.3242,
+      "step": 1323
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3867332882698016e-05,
+      "loss": 1.2828,
+      "step": 1324
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.385751060412235e-05,
+      "loss": 1.2589,
+      "step": 1325
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3847681892825865e-05,
+      "loss": 1.287,
+      "step": 1326
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.383784675337108e-05,
+      "loss": 1.3118,
+      "step": 1327
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.382800519032353e-05,
+      "loss": 1.3099,
+      "step": 1328
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.38181572082517e-05,
+      "loss": 1.2927,
+      "step": 1329
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3808302811727074e-05,
+      "loss": 1.3071,
+      "step": 1330
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.379844200532411e-05,
+      "loss": 1.3211,
+      "step": 1331
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.378857479362024e-05,
+      "loss": 1.3449,
+      "step": 1332
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.377870118119587e-05,
+      "loss": 1.2573,
+      "step": 1333
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.376882117263437e-05,
+      "loss": 1.3084,
+      "step": 1334
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.37589347725221e-05,
+      "loss": 1.2562,
+      "step": 1335
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.374904198544836e-05,
+      "loss": 1.3295,
+      "step": 1336
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.373914281600544e-05,
+      "loss": 1.3231,
+      "step": 1337
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.372923726878856e-05,
+      "loss": 1.3315,
+      "step": 1338
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.371932534839594e-05,
+      "loss": 1.3093,
+      "step": 1339
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.370940705942874e-05,
+      "loss": 1.2672,
+      "step": 1340
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.369948240649106e-05,
+      "loss": 1.3012,
+      "step": 1341
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.368955139418998e-05,
+      "loss": 1.3134,
+      "step": 1342
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.367961402713553e-05,
+      "loss": 1.3119,
+      "step": 1343
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3669670309940663e-05,
+      "loss": 1.2915,
+      "step": 1344
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.365972024722131e-05,
+      "loss": 1.2518,
+      "step": 1345
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3649763843596334e-05,
+      "loss": 1.2958,
+      "step": 1346
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.363980110368755e-05,
+      "loss": 1.2964,
+      "step": 1347
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.36298320321197e-05,
+      "loss": 1.333,
+      "step": 1348
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.361985663352048e-05,
+      "loss": 1.3069,
+      "step": 1349
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.360987491252051e-05,
+      "loss": 1.2551,
+      "step": 1350
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.359988687375336e-05,
+      "loss": 1.2971,
+      "step": 1351
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3589892521855515e-05,
+      "loss": 1.3038,
+      "step": 1352
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.35798918614664e-05,
+      "loss": 1.3001,
+      "step": 1353
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.356988489722837e-05,
+      "loss": 1.3387,
+      "step": 1354
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.355987163378671e-05,
+      "loss": 1.2484,
+      "step": 1355
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.354985207578961e-05,
+      "loss": 1.2616,
+      "step": 1356
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3539826227888216e-05,
+      "loss": 1.3249,
+      "step": 1357
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.352979409473656e-05,
+      "loss": 1.2968,
+      "step": 1358
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.351975568099159e-05,
+      "loss": 1.3149,
+      "step": 1359
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.350971099131322e-05,
+      "loss": 1.362,
+      "step": 1360
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.349966003036421e-05,
+      "loss": 1.2973,
+      "step": 1361
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3489602802810276e-05,
+      "loss": 1.2954,
+      "step": 1362
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.347953931332004e-05,
+      "loss": 1.2829,
+      "step": 1363
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3469469566565e-05,
+      "loss": 1.2816,
+      "step": 1364
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.345939356721959e-05,
+      "loss": 1.3187,
+      "step": 1365
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3449311319961134e-05,
+      "loss": 1.2707,
+      "step": 1366
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.343922282946985e-05,
+      "loss": 1.3621,
+      "step": 1367
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.342912810042888e-05,
+      "loss": 1.3427,
+      "step": 1368
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3419027137524236e-05,
+      "loss": 1.271,
+      "step": 1369
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.340891994544483e-05,
+      "loss": 1.2508,
+      "step": 1370
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.339880652888246e-05,
+      "loss": 1.2675,
+      "step": 1371
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.338868689253183e-05,
+      "loss": 1.286,
+      "step": 1372
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.337856104109053e-05,
+      "loss": 1.2766,
+      "step": 1373
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3368428979259006e-05,
+      "loss": 1.2772,
+      "step": 1374
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.335829071174063e-05,
+      "loss": 1.3164,
+      "step": 1375
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.334814624324163e-05,
+      "loss": 1.2738,
+      "step": 1376
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.333799557847109e-05,
+      "loss": 1.3249,
+      "step": 1377
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.332783872214103e-05,
+      "loss": 1.3157,
+      "step": 1378
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.331767567896629e-05,
+      "loss": 1.2825,
+      "step": 1379
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.330750645366461e-05,
+      "loss": 1.2947,
+      "step": 1380
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3297331050956576e-05,
+      "loss": 1.3257,
+      "step": 1381
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.328714947556568e-05,
+      "loss": 1.2804,
+      "step": 1382
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.327696173221824e-05,
+      "loss": 1.2964,
+      "step": 1383
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.326676782564347e-05,
+      "loss": 1.3306,
+      "step": 1384
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.325656776057341e-05,
+      "loss": 1.2447,
+      "step": 1385
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.324636154174299e-05,
+      "loss": 1.3169,
+      "step": 1386
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3236149173889975e-05,
+      "loss": 1.2883,
+      "step": 1387
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3225930661755005e-05,
+      "loss": 1.3379,
+      "step": 1388
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.321570601008155e-05,
+      "loss": 1.3591,
+      "step": 1389
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.320547522361595e-05,
+      "loss": 1.3079,
+      "step": 1390
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3195238307107375e-05,
+      "loss": 1.3126,
+      "step": 1391
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.318499526530786e-05,
+      "loss": 1.3291,
+      "step": 1392
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.317474610297226e-05,
+      "loss": 1.3011,
+      "step": 1393
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.316449082485829e-05,
+      "loss": 1.302,
+      "step": 1394
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.31542294357265e-05,
+      "loss": 1.3377,
+      "step": 1395
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3143961940340274e-05,
+      "loss": 1.2829,
+      "step": 1396
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.313368834346583e-05,
+      "loss": 1.2995,
+      "step": 1397
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.312340864987221e-05,
+      "loss": 1.3037,
+      "step": 1398
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.311312286433131e-05,
+      "loss": 1.3276,
+      "step": 1399
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.310283099161783e-05,
+      "loss": 1.3256,
+      "step": 1400
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.309253303650932e-05,
+      "loss": 1.3359,
+      "step": 1401
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3082229003786114e-05,
+      "loss": 1.3255,
+      "step": 1402
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.3071918898231413e-05,
+      "loss": 1.2528,
+      "step": 1403
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.3061602724631205e-05,
+      "loss": 1.3548,
+      "step": 1404
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.3051280487774316e-05,
+      "loss": 1.2952,
+      "step": 1405
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.304095219245236e-05,
+      "loss": 1.3009,
+      "step": 1406
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.303061784345979e-05,
+      "loss": 1.3164,
+      "step": 1407
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.302027744559387e-05,
+      "loss": 1.2863,
+      "step": 1408
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.300993100365463e-05,
+      "loss": 1.2838,
+      "step": 1409
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.299957852244496e-05,
+      "loss": 1.3443,
+      "step": 1410
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.298922000677053e-05,
+      "loss": 1.278,
+      "step": 1411
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2978855461439806e-05,
+      "loss": 1.2888,
+      "step": 1412
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.296848489126406e-05,
+      "loss": 1.2988,
+      "step": 1413
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.295810830105736e-05,
+      "loss": 1.276,
+      "step": 1414
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.294772569563656e-05,
+      "loss": 1.2776,
+      "step": 1415
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.293733707982132e-05,
+      "loss": 1.2864,
+      "step": 1416
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.292694245843407e-05,
+      "loss": 1.2906,
+      "step": 1417
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2916541836300065e-05,
+      "loss": 1.3245,
+      "step": 1418
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.290613521824731e-05,
+      "loss": 1.3229,
+      "step": 1419
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.28957226091066e-05,
+      "loss": 1.2939,
+      "step": 1420
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2885304013711525e-05,
+      "loss": 1.2817,
+      "step": 1421
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2874879436898444e-05,
+      "loss": 1.2686,
+      "step": 1422
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.286444888350649e-05,
+      "loss": 1.3233,
+      "step": 1423
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.285401235837758e-05,
+      "loss": 1.2612,
+      "step": 1424
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.28435698663564e-05,
+      "loss": 1.3166,
+      "step": 1425
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.283312141229039e-05,
+      "loss": 1.3319,
+      "step": 1426
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.282266700102978e-05,
+      "loss": 1.2886,
+      "step": 1427
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.281220663742756e-05,
+      "loss": 1.237,
+      "step": 1428
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.280174032633947e-05,
+      "loss": 1.3069,
+      "step": 1429
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.279126807262403e-05,
+      "loss": 1.2735,
+      "step": 1430
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.27807898811425e-05,
+      "loss": 1.2644,
+      "step": 1431
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.277030575675891e-05,
+      "loss": 1.2884,
+      "step": 1432
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2759815704340034e-05,
+      "loss": 1.2894,
+      "step": 1433
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2749319728755415e-05,
+      "loss": 1.2695,
+      "step": 1434
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.273881783487732e-05,
+      "loss": 1.3131,
+      "step": 1435
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2728310027580786e-05,
+      "loss": 1.3413,
+      "step": 1436
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.271779631174358e-05,
+      "loss": 1.2963,
+      "step": 1437
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.270727669224622e-05,
+      "loss": 1.3176,
+      "step": 1438
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.269675117397196e-05,
+      "loss": 1.3155,
+      "step": 1439
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.268621976180681e-05,
+      "loss": 1.2833,
+      "step": 1440
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.267568246063948e-05,
+      "loss": 1.2208,
+      "step": 1441
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2665139275361446e-05,
+      "loss": 1.3166,
+      "step": 1442
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.26545902108669e-05,
+      "loss": 1.2933,
+      "step": 1443
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2644035272052756e-05,
+      "loss": 1.3392,
+      "step": 1444
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.263347446381869e-05,
+      "loss": 1.2914,
+      "step": 1445
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2622907791067056e-05,
+      "loss": 1.2961,
+      "step": 1446
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.261233525870296e-05,
+      "loss": 1.3064,
+      "step": 1447
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.260175687163423e-05,
+      "loss": 1.3228,
+      "step": 1448
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.259117263477138e-05,
+      "loss": 1.2475,
+      "step": 1449
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2580582553027684e-05,
+      "loss": 1.2894,
+      "step": 1450
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2569986631319104e-05,
+      "loss": 1.2327,
+      "step": 1451
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2559384874564305e-05,
+      "loss": 1.2908,
+      "step": 1452
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.254877728768468e-05,
+      "loss": 1.2182,
+      "step": 1453
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2538163875604316e-05,
+      "loss": 1.2867,
+      "step": 1454
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.252754464325001e-05,
+      "loss": 1.2964,
+      "step": 1455
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.251691959555126e-05,
+      "loss": 1.299,
+      "step": 1456
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2506288737440265e-05,
+      "loss": 1.2994,
+      "step": 1457
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.249565207385192e-05,
+      "loss": 1.3188,
+      "step": 1458
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.24850096097238e-05,
+      "loss": 1.2994,
+      "step": 1459
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2474361349996205e-05,
+      "loss": 1.3603,
+      "step": 1460
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2463707299612086e-05,
+      "loss": 1.3068,
+      "step": 1461
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.245304746351712e-05,
+      "loss": 1.3158,
+      "step": 1462
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2442381846659644e-05,
+      "loss": 1.3028,
+      "step": 1463
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.243171045399069e-05,
+      "loss": 1.3344,
+      "step": 1464
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2421033290463966e-05,
+      "loss": 1.262,
+      "step": 1465
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.241035036103587e-05,
+      "loss": 1.2577,
+      "step": 1466
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.239966167066545e-05,
+      "loss": 1.2835,
+      "step": 1467
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2388967224314464e-05,
+      "loss": 1.2941,
+      "step": 1468
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2378267026947314e-05,
+      "loss": 1.2777,
+      "step": 1469
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2367561083531074e-05,
+      "loss": 1.2969,
+      "step": 1470
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.235684939903551e-05,
+      "loss": 1.3229,
+      "step": 1471
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.234613197843302e-05,
+      "loss": 1.31,
+      "step": 1472
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.233540882669869e-05,
+      "loss": 1.2708,
+      "step": 1473
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.232467994881026e-05,
+      "loss": 1.2625,
+      "step": 1474
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2313945349748116e-05,
+      "loss": 1.3055,
+      "step": 1475
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.230320503449531e-05,
+      "loss": 1.2882,
+      "step": 1476
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2292459008037554e-05,
+      "loss": 1.3261,
+      "step": 1477
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.228170727536319e-05,
+      "loss": 1.3127,
+      "step": 1478
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.227094984146323e-05,
+      "loss": 1.2909,
+      "step": 1479
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.226018671133134e-05,
+      "loss": 1.2853,
+      "step": 1480
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.224941788996378e-05,
+      "loss": 1.2684,
+      "step": 1481
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.223864338235951e-05,
+      "loss": 1.3026,
+      "step": 1482
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2227863193520115e-05,
+      "loss": 1.3298,
+      "step": 1483
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.221707732844979e-05,
+      "loss": 1.2929,
+      "step": 1484
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.220628579215539e-05,
+      "loss": 1.3084,
+      "step": 1485
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.21954885896464e-05,
+      "loss": 1.3481,
+      "step": 1486
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2184685725934926e-05,
+      "loss": 1.286,
+      "step": 1487
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2173877206035714e-05,
+      "loss": 1.3556,
+      "step": 1488
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2163063034966126e-05,
+      "loss": 1.2548,
+      "step": 1489
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.215224321774614e-05,
+      "loss": 1.2961,
+      "step": 1490
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.214141775939839e-05,
+      "loss": 1.2892,
+      "step": 1491
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.213058666494807e-05,
+      "loss": 1.2803,
+      "step": 1492
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2119749939423063e-05,
+      "loss": 1.3108,
+      "step": 1493
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2108907587853794e-05,
+      "loss": 1.3067,
+      "step": 1494
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2098059615273354e-05,
+      "loss": 1.2733,
+      "step": 1495
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2087206026717415e-05,
+      "loss": 1.2772,
+      "step": 1496
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.207634682722427e-05,
+      "loss": 1.2983,
+      "step": 1497
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.206548202183479e-05,
+      "loss": 1.297,
+      "step": 1498
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.20546116155925e-05,
+      "loss": 1.2824,
+      "step": 1499
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.204373561354345e-05,
+      "loss": 1.2834,
+      "step": 1500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.203285402073637e-05,
+      "loss": 1.3461,
+      "step": 1501
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.202196684222253e-05,
+      "loss": 1.2753,
+      "step": 1502
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.2011074083055814e-05,
+      "loss": 1.283,
+      "step": 1503
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.200017574829268e-05,
+      "loss": 1.3074,
+      "step": 1504
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.19892718429922e-05,
+      "loss": 1.2996,
+      "step": 1505
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1978362372216006e-05,
+      "loss": 1.3166,
+      "step": 1506
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.196744734102833e-05,
+      "loss": 1.3329,
+      "step": 1507
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.195652675449599e-05,
+      "loss": 1.2842,
+      "step": 1508
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.194560061768835e-05,
+      "loss": 1.314,
+      "step": 1509
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.193466893567739e-05,
+      "loss": 1.2629,
+      "step": 1510
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.192373171353765e-05,
+      "loss": 1.3123,
+      "step": 1511
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1912788956346226e-05,
+      "loss": 1.2651,
+      "step": 1512
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.190184066918281e-05,
+      "loss": 1.2869,
+      "step": 1513
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.189088685712964e-05,
+      "loss": 1.3235,
+      "step": 1514
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.187992752527153e-05,
+      "loss": 1.2989,
+      "step": 1515
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.186896267869585e-05,
+      "loss": 1.3157,
+      "step": 1516
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1857992322492525e-05,
+      "loss": 1.308,
+      "step": 1517
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.184701646175407e-05,
+      "loss": 1.2973,
+      "step": 1518
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.183603510157551e-05,
+      "loss": 1.3628,
+      "step": 1519
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1825048247054444e-05,
+      "loss": 1.3117,
+      "step": 1520
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.181405590329102e-05,
+      "loss": 1.2541,
+      "step": 1521
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.180305807538796e-05,
+      "loss": 1.3173,
+      "step": 1522
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1792054768450466e-05,
+      "loss": 1.267,
+      "step": 1523
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.178104598758636e-05,
+      "loss": 1.3344,
+      "step": 1524
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1770031737905946e-05,
+      "loss": 1.3157,
+      "step": 1525
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1759012024522103e-05,
+      "loss": 1.2994,
+      "step": 1526
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1747986852550225e-05,
+      "loss": 1.3523,
+      "step": 1527
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.173695622710826e-05,
+      "loss": 1.3186,
+      "step": 1528
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.172592015331666e-05,
+      "loss": 1.2854,
+      "step": 1529
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.171487863629843e-05,
+      "loss": 1.2563,
+      "step": 1530
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1703831681179096e-05,
+      "loss": 1.3092,
+      "step": 1531
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.169277929308669e-05,
+      "loss": 1.3017,
+      "step": 1532
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.168172147715181e-05,
+      "loss": 1.299,
+      "step": 1533
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1670658238507524e-05,
+      "loss": 1.3178,
+      "step": 1534
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1659589582289446e-05,
+      "loss": 1.2963,
+      "step": 1535
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.16485155136357e-05,
+      "loss": 1.3274,
+      "step": 1536
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.16374360376869e-05,
+      "loss": 1.2938,
+      "step": 1537
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1626351159586224e-05,
+      "loss": 1.2672,
+      "step": 1538
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1615260884479304e-05,
+      "loss": 1.3287,
+      "step": 1539
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1604165217514296e-05,
+      "loss": 1.3223,
+      "step": 1540
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.159306416384187e-05,
+      "loss": 1.2831,
+      "step": 1541
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.158195772861517e-05,
+      "loss": 1.3246,
+      "step": 1542
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1570845916989875e-05,
+      "loss": 1.3066,
+      "step": 1543
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1559728734124125e-05,
+      "loss": 1.3554,
+      "step": 1544
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.154860618517858e-05,
+      "loss": 1.2883,
+      "step": 1545
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1537478275316364e-05,
+      "loss": 1.2856,
+      "step": 1546
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.152634500970312e-05,
+      "loss": 1.3114,
+      "step": 1547
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.151520639350695e-05,
+      "loss": 1.3208,
+      "step": 1548
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.150406243189846e-05,
+      "loss": 1.3286,
+      "step": 1549
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1492913130050715e-05,
+      "loss": 1.3066,
+      "step": 1550
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1481758493139295e-05,
+      "loss": 1.3445,
+      "step": 1551
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.147059852634221e-05,
+      "loss": 1.3027,
+      "step": 1552
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.145943323483999e-05,
+      "loss": 1.3306,
+      "step": 1553
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.144826262381559e-05,
+      "loss": 1.3253,
+      "step": 1554
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.143708669845449e-05,
+      "loss": 1.2997,
+      "step": 1555
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.142590546394459e-05,
+      "loss": 1.2951,
+      "step": 1556
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.141471892547627e-05,
+      "loss": 1.3165,
+      "step": 1557
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.140352708824237e-05,
+      "loss": 1.2899,
+      "step": 1558
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1392329957438203e-05,
+      "loss": 1.3088,
+      "step": 1559
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1381127538261524e-05,
+      "loss": 1.298,
+      "step": 1560
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.136991983591255e-05,
+      "loss": 1.2857,
+      "step": 1561
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1358706855593935e-05,
+      "loss": 1.2902,
+      "step": 1562
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1347488602510824e-05,
+      "loss": 1.3583,
+      "step": 1563
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.133626508187076e-05,
+      "loss": 1.3131,
+      "step": 1564
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.132503629888376e-05,
+      "loss": 1.2533,
+      "step": 1565
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.131380225876228e-05,
+      "loss": 1.2924,
+      "step": 1566
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.130256296672121e-05,
+      "loss": 1.3152,
+      "step": 1567
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1291318427977876e-05,
+      "loss": 1.2727,
+      "step": 1568
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1280068647752066e-05,
+      "loss": 1.3052,
+      "step": 1569
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.126881363126595e-05,
+      "loss": 1.3038,
+      "step": 1570
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1257553383744186e-05,
+      "loss": 1.3172,
+      "step": 1571
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1246287910413824e-05,
+      "loss": 1.2669,
+      "step": 1572
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.123501721650434e-05,
+      "loss": 1.2939,
+      "step": 1573
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.122374130724765e-05,
+      "loss": 1.302,
+      "step": 1574
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1212460187878085e-05,
+      "loss": 1.2917,
+      "step": 1575
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1201173863632396e-05,
+      "loss": 1.3215,
+      "step": 1576
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1189882339749735e-05,
+      "loss": 1.281,
+      "step": 1577
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.117858562147169e-05,
+      "loss": 1.3307,
+      "step": 1578
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.116728371404225e-05,
+      "loss": 1.2713,
+      "step": 1579
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.115597662270781e-05,
+      "loss": 1.2681,
+      "step": 1580
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.114466435271717e-05,
+      "loss": 1.2676,
+      "step": 1581
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.113334690932155e-05,
+      "loss": 1.2849,
+      "step": 1582
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1122024297774545e-05,
+      "loss": 1.3106,
+      "step": 1583
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.111069652333219e-05,
+      "loss": 1.3076,
+      "step": 1584
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1099363591252844e-05,
+      "loss": 1.2546,
+      "step": 1585
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1088025506797356e-05,
+      "loss": 1.3116,
+      "step": 1586
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.107668227522889e-05,
+      "loss": 1.3047,
+      "step": 1587
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.106533390181304e-05,
+      "loss": 1.3274,
+      "step": 1588
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.105398039181775e-05,
+      "loss": 1.2934,
+      "step": 1589
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1042621750513405e-05,
+      "loss": 1.2906,
+      "step": 1590
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.103125798317272e-05,
+      "loss": 1.26,
+      "step": 1591
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.101988909507081e-05,
+      "loss": 1.2634,
+      "step": 1592
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.100851509148517e-05,
+      "loss": 1.2759,
+      "step": 1593
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.099713597769566e-05,
+      "loss": 1.3192,
+      "step": 1594
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.098575175898452e-05,
+      "loss": 1.2767,
+      "step": 1595
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.097436244063636e-05,
+      "loss": 1.2558,
+      "step": 1596
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0962968027938156e-05,
+      "loss": 1.3523,
+      "step": 1597
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0951568526179235e-05,
+      "loss": 1.3043,
+      "step": 1598
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.094016394065131e-05,
+      "loss": 1.3225,
+      "step": 1599
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0928754276648443e-05,
+      "loss": 1.3094,
+      "step": 1600
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.091733953946705e-05,
+      "loss": 1.2396,
+      "step": 1601
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.09059197344059e-05,
+      "loss": 1.2891,
+      "step": 1602
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.089449486676613e-05,
+      "loss": 1.3109,
+      "step": 1603
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.08830649418512e-05,
+      "loss": 1.3257,
+      "step": 1604
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.087162996496696e-05,
+      "loss": 1.3005,
+      "step": 1605
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.086018994142156e-05,
+      "loss": 1.3104,
+      "step": 1606
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.084874487652551e-05,
+      "loss": 1.3131,
+      "step": 1607
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0837294775591675e-05,
+      "loss": 1.3353,
+      "step": 1608
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.082583964393524e-05,
+      "loss": 1.3006,
+      "step": 1609
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.081437948687373e-05,
+      "loss": 1.3282,
+      "step": 1610
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0802914309727004e-05,
+      "loss": 1.3562,
+      "step": 1611
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0791444117817247e-05,
+      "loss": 1.2589,
+      "step": 1612
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0779968916468974e-05,
+      "loss": 1.3137,
+      "step": 1613
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.076848871100904e-05,
+      "loss": 1.3063,
+      "step": 1614
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.075700350676659e-05,
+      "loss": 1.3101,
+      "step": 1615
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.074551330907312e-05,
+      "loss": 1.3169,
+      "step": 1616
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.073401812326244e-05,
+      "loss": 1.2861,
+      "step": 1617
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.072251795467065e-05,
+      "loss": 1.306,
+      "step": 1618
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.071101280863621e-05,
+      "loss": 1.2439,
+      "step": 1619
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.069950269049983e-05,
+      "loss": 1.289,
+      "step": 1620
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.068798760560458e-05,
+      "loss": 1.3182,
+      "step": 1621
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.067646755929582e-05,
+      "loss": 1.2638,
+      "step": 1622
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.066494255692119e-05,
+      "loss": 1.2697,
+      "step": 1623
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0653412603830665e-05,
+      "loss": 1.3101,
+      "step": 1624
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.06418777053765e-05,
+      "loss": 1.3043,
+      "step": 1625
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0630337866913236e-05,
+      "loss": 1.2993,
+      "step": 1626
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.061879309379774e-05,
+      "loss": 1.2612,
+      "step": 1627
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.060724339138913e-05,
+      "loss": 1.3094,
+      "step": 1628
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0595688765048855e-05,
+      "loss": 1.2825,
+      "step": 1629
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.058412922014061e-05,
+      "loss": 1.353,
+      "step": 1630
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.057256476203038e-05,
+      "loss": 1.2747,
+      "step": 1631
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.056099539608646e-05,
+      "loss": 1.3117,
+      "step": 1632
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0549421127679395e-05,
+      "loss": 1.3436,
+      "step": 1633
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.053784196218201e-05,
+      "loss": 1.3072,
+      "step": 1634
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.052625790496942e-05,
+      "loss": 1.3141,
+      "step": 1635
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0514668961418984e-05,
+      "loss": 1.2224,
+      "step": 1636
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.050307513691035e-05,
+      "loss": 1.2675,
+      "step": 1637
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0491476436825427e-05,
+      "loss": 1.3372,
+      "step": 1638
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.047987286654838e-05,
+      "loss": 1.2686,
+      "step": 1639
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0468264431465643e-05,
+      "loss": 1.3408,
+      "step": 1640
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.045665113696591e-05,
+      "loss": 1.3472,
+      "step": 1641
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0445032988440126e-05,
+      "loss": 1.301,
+      "step": 1642
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0433409991281483e-05,
+      "loss": 1.2907,
+      "step": 1643
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.042178215088543e-05,
+      "loss": 1.2662,
+      "step": 1644
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.041014947264967e-05,
+      "loss": 1.239,
+      "step": 1645
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0398511961974143e-05,
+      "loss": 1.2055,
+      "step": 1646
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0386869624261036e-05,
+      "loss": 1.2929,
+      "step": 1647
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0375222464914782e-05,
+      "loss": 1.2832,
+      "step": 1648
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0363570489342033e-05,
+      "loss": 1.2852,
+      "step": 1649
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.03519137029517e-05,
+      "loss": 1.2865,
+      "step": 1650
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.034025211115492e-05,
+      "loss": 1.2482,
+      "step": 1651
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0328585719365057e-05,
+      "loss": 1.2963,
+      "step": 1652
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0316914532997694e-05,
+      "loss": 1.2997,
+      "step": 1653
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.030523855747066e-05,
+      "loss": 1.3042,
+      "step": 1654
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0293557798203998e-05,
+      "loss": 1.3074,
+      "step": 1655
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0281872260619965e-05,
+      "loss": 1.3307,
+      "step": 1656
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0270181950143045e-05,
+      "loss": 1.2965,
+      "step": 1657
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.025848687219993e-05,
+      "loss": 1.3051,
+      "step": 1658
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0246787032219535e-05,
+      "loss": 1.344,
+      "step": 1659
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0235082435632984e-05,
+      "loss": 1.3226,
+      "step": 1660
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.022337308787359e-05,
+      "loss": 1.2981,
+      "step": 1661
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.02116589943769e-05,
+      "loss": 1.2949,
+      "step": 1662
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.019994016058064e-05,
+      "loss": 1.2734,
+      "step": 1663
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.018821659192476e-05,
+      "loss": 1.2922,
+      "step": 1664
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0176488293851388e-05,
+      "loss": 1.2911,
+      "step": 1665
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0164755271804856e-05,
+      "loss": 1.2955,
+      "step": 1666
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.015301753123169e-05,
+      "loss": 1.327,
+      "step": 1667
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0141275077580592e-05,
+      "loss": 1.2304,
+      "step": 1668
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0129527916302482e-05,
+      "loss": 1.3248,
+      "step": 1669
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0117776052850427e-05,
+      "loss": 1.2967,
+      "step": 1670
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0106019492679714e-05,
+      "loss": 1.2503,
+      "step": 1671
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.009425824124778e-05,
+      "loss": 1.2552,
+      "step": 1672
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.008249230401426e-05,
+      "loss": 1.2909,
+      "step": 1673
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0070721686440953e-05,
+      "loss": 1.3333,
+      "step": 1674
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0058946393991833e-05,
+      "loss": 1.2823,
+      "step": 1675
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.004716643213305e-05,
+      "loss": 1.2726,
+      "step": 1676
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.003538180633292e-05,
+      "loss": 1.3186,
+      "step": 1677
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0023592522061916e-05,
+      "loss": 1.2841,
+      "step": 1678
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0011798584792672e-05,
+      "loss": 1.3068,
+      "step": 1679
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0000000000000004e-05,
+      "loss": 1.2998,
+      "step": 1680
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9988196773160857e-05,
+      "loss": 1.3281,
+      "step": 1681
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9976388909754348e-05,
+      "loss": 1.2653,
+      "step": 1682
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.996457641526174e-05,
+      "loss": 1.2684,
+      "step": 1683
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.995275929516646e-05,
+      "loss": 1.3073,
+      "step": 1684
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9940937554954053e-05,
+      "loss": 1.287,
+      "step": 1685
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9929111200112233e-05,
+      "loss": 1.2919,
+      "step": 1686
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.991728023613085e-05,
+      "loss": 1.2492,
+      "step": 1687
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.990544466850189e-05,
+      "loss": 1.2469,
+      "step": 1688
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9893604502719474e-05,
+      "loss": 1.3062,
+      "step": 1689
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9881759744279875e-05,
+      "loss": 1.326,
+      "step": 1690
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.986991039868148e-05,
+      "loss": 1.2793,
+      "step": 1691
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9858056471424804e-05,
+      "loss": 1.3325,
+      "step": 1692
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.98461979680125e-05,
+      "loss": 1.306,
+      "step": 1693
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.983433489394934e-05,
+      "loss": 1.334,
+      "step": 1694
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9822467254742212e-05,
+      "loss": 1.3122,
+      "step": 1695
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9810595055900148e-05,
+      "loss": 1.2584,
+      "step": 1696
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9798718302934255e-05,
+      "loss": 1.3124,
+      "step": 1697
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9786837001357782e-05,
+      "loss": 1.2975,
+      "step": 1698
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9774951156686094e-05,
+      "loss": 1.3272,
+      "step": 1699
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.976306077443665e-05,
+      "loss": 1.2671,
+      "step": 1700
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9751165860129024e-05,
+      "loss": 1.305,
+      "step": 1701
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.973926641928489e-05,
+      "loss": 1.3157,
+      "step": 1702
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9727362457428012e-05,
+      "loss": 1.289,
+      "step": 1703
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.971545398008428e-05,
+      "loss": 1.265,
+      "step": 1704
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.970354099278166e-05,
+      "loss": 1.2552,
+      "step": 1705
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9691623501050212e-05,
+      "loss": 1.2669,
+      "step": 1706
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.967970151042209e-05,
+      "loss": 1.3045,
+      "step": 1707
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9667775026431544e-05,
+      "loss": 1.323,
+      "step": 1708
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.96558440546149e-05,
+      "loss": 1.2951,
+      "step": 1709
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9643908600510572e-05,
+      "loss": 1.3034,
+      "step": 1710
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9631968669659047e-05,
+      "loss": 1.3261,
+      "step": 1711
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9620024267602906e-05,
+      "loss": 1.2854,
+      "step": 1712
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9608075399886792e-05,
+      "loss": 1.3118,
+      "step": 1713
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9596122072057424e-05,
+      "loss": 1.2566,
+      "step": 1714
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.958416428966359e-05,
+      "loss": 1.2837,
+      "step": 1715
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.957220205825615e-05,
+      "loss": 1.2366,
+      "step": 1716
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.956023538338803e-05,
+      "loss": 1.3016,
+      "step": 1717
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9548264270614217e-05,
+      "loss": 1.265,
+      "step": 1718
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.953628872549175e-05,
+      "loss": 1.2633,
+      "step": 1719
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.952430875357974e-05,
+      "loss": 1.2747,
+      "step": 1720
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9512324360439347e-05,
+      "loss": 1.279,
+      "step": 1721
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9500335551633773e-05,
+      "loss": 1.3096,
+      "step": 1722
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9488342332728276e-05,
+      "loss": 1.3025,
+      "step": 1723
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9476344709290175e-05,
+      "loss": 1.2609,
+      "step": 1724
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9464342686888826e-05,
+      "loss": 1.2991,
+      "step": 1725
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9452336271095613e-05,
+      "loss": 1.2917,
+      "step": 1726
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9440325467483974e-05,
+      "loss": 1.3211,
+      "step": 1727
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.942831028162938e-05,
+      "loss": 1.3183,
+      "step": 1728
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9416290719109333e-05,
+      "loss": 1.3264,
+      "step": 1729
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9404266785503376e-05,
+      "loss": 1.2962,
+      "step": 1730
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9392238486393068e-05,
+      "loss": 1.28,
+      "step": 1731
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9380205827362007e-05,
+      "loss": 1.2701,
+      "step": 1732
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9368168813995806e-05,
+      "loss": 1.2984,
+      "step": 1733
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9356127451882105e-05,
+      "loss": 1.2856,
+      "step": 1734
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.934408174661055e-05,
+      "loss": 1.264,
+      "step": 1735
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.933203170377283e-05,
+      "loss": 1.2299,
+      "step": 1736
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.931997732896262e-05,
+      "loss": 1.2662,
+      "step": 1737
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9307918627775627e-05,
+      "loss": 1.2906,
+      "step": 1738
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9295855605809543e-05,
+      "loss": 1.2641,
+      "step": 1739
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9283788268664085e-05,
+      "loss": 1.3194,
+      "step": 1740
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9271716621940965e-05,
+      "loss": 1.3055,
+      "step": 1741
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9259640671243903e-05,
+      "loss": 1.3295,
+      "step": 1742
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9247560422178604e-05,
+      "loss": 1.3002,
+      "step": 1743
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.923547588035278e-05,
+      "loss": 1.2909,
+      "step": 1744
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9223387051376133e-05,
+      "loss": 1.3057,
+      "step": 1745
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.921129394086035e-05,
+      "loss": 1.3612,
+      "step": 1746
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.919919655441911e-05,
+      "loss": 1.2866,
+      "step": 1747
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9187094897668076e-05,
+      "loss": 1.277,
+      "step": 1748
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9174988976224897e-05,
+      "loss": 1.241,
+      "step": 1749
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.916287879570919e-05,
+      "loss": 1.2827,
+      "step": 1750
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9150764361742564e-05,
+      "loss": 1.3067,
+      "step": 1751
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9138645679948596e-05,
+      "loss": 1.3002,
+      "step": 1752
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.912652275595283e-05,
+      "loss": 1.3055,
+      "step": 1753
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9114395595382786e-05,
+      "loss": 1.2862,
+      "step": 1754
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9102264203867948e-05,
+      "loss": 1.284,
+      "step": 1755
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9090128587039763e-05,
+      "loss": 1.3091,
+      "step": 1756
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9077988750531645e-05,
+      "loss": 1.2661,
+      "step": 1757
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9065844699978957e-05,
+      "loss": 1.3209,
+      "step": 1758
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9053696441019022e-05,
+      "loss": 1.2794,
+      "step": 1759
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9041543979291125e-05,
+      "loss": 1.2528,
+      "step": 1760
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.90293873204365e-05,
+      "loss": 1.2654,
+      "step": 1761
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9017226470098307e-05,
+      "loss": 1.2904,
+      "step": 1762
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9005061433921685e-05,
+      "loss": 1.2665,
+      "step": 1763
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8992892217553693e-05,
+      "loss": 1.3105,
+      "step": 1764
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.898071882664334e-05,
+      "loss": 1.2745,
+      "step": 1765
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.896854126684157e-05,
+      "loss": 1.2836,
+      "step": 1766
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.895635954380127e-05,
+      "loss": 1.2484,
+      "step": 1767
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.894417366317724e-05,
+      "loss": 1.2963,
+      "step": 1768
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8931983630626218e-05,
+      "loss": 1.2534,
+      "step": 1769
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8919789451806893e-05,
+      "loss": 1.285,
+      "step": 1770
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.890759113237985e-05,
+      "loss": 1.2559,
+      "step": 1771
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8895388678007602e-05,
+      "loss": 1.3274,
+      "step": 1772
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8883182094354594e-05,
+      "loss": 1.1933,
+      "step": 1773
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.887097138708717e-05,
+      "loss": 1.266,
+      "step": 1774
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8858756561873605e-05,
+      "loss": 1.2943,
+      "step": 1775
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.884653762438407e-05,
+      "loss": 1.3024,
+      "step": 1776
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8834314580290655e-05,
+      "loss": 1.2913,
+      "step": 1777
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.882208743526736e-05,
+      "loss": 1.2376,
+      "step": 1778
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8809856194990067e-05,
+      "loss": 1.3143,
+      "step": 1779
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8797620865136594e-05,
+      "loss": 1.2916,
+      "step": 1780
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8785381451386628e-05,
+      "loss": 1.3028,
+      "step": 1781
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.877313795942176e-05,
+      "loss": 1.2968,
+      "step": 1782
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8760890394925477e-05,
+      "loss": 1.2332,
+      "step": 1783
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8748638763583158e-05,
+      "loss": 1.3258,
+      "step": 1784
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8736383071082065e-05,
+      "loss": 1.3252,
+      "step": 1785
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.872412332311135e-05,
+      "loss": 1.2971,
+      "step": 1786
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8711859525362045e-05,
+      "loss": 1.2805,
+      "step": 1787
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8699591683527058e-05,
+      "loss": 1.3081,
+      "step": 1788
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8687319803301183e-05,
+      "loss": 1.2288,
+      "step": 1789
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.867504389038108e-05,
+      "loss": 1.2787,
+      "step": 1790
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8662763950465284e-05,
+      "loss": 1.3432,
+      "step": 1791
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8650479989254206e-05,
+      "loss": 1.2632,
+      "step": 1792
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8638192012450115e-05,
+      "loss": 1.266,
+      "step": 1793
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.862590002575714e-05,
+      "loss": 1.2915,
+      "step": 1794
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.861360403488129e-05,
+      "loss": 1.3058,
+      "step": 1795
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8601304045530414e-05,
+      "loss": 1.2763,
+      "step": 1796
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.858900006341422e-05,
+      "loss": 1.2801,
+      "step": 1797
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8576692094244286e-05,
+      "loss": 1.2901,
+      "step": 1798
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.856438014373402e-05,
+      "loss": 1.2949,
+      "step": 1799
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8552064217598697e-05,
+      "loss": 1.2914,
+      "step": 1800
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.853974432155541e-05,
+      "loss": 1.3045,
+      "step": 1801
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.852742046132312e-05,
+      "loss": 1.227,
+      "step": 1802
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8515092642622625e-05,
+      "loss": 1.2859,
+      "step": 1803
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8502760871176546e-05,
+      "loss": 1.2837,
+      "step": 1804
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8490425152709367e-05,
+      "loss": 1.3576,
+      "step": 1805
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.847808549294736e-05,
+      "loss": 1.2636,
+      "step": 1806
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8465741897618673e-05,
+      "loss": 1.3152,
+      "step": 1807
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8453394372453253e-05,
+      "loss": 1.348,
+      "step": 1808
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8441042923182872e-05,
+      "loss": 1.2609,
+      "step": 1809
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.842868755554114e-05,
+      "loss": 1.2705,
+      "step": 1810
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8416328275263472e-05,
+      "loss": 1.2899,
+      "step": 1811
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8403965088087105e-05,
+      "loss": 1.3033,
+      "step": 1812
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.839159799975109e-05,
+      "loss": 1.3166,
+      "step": 1813
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8379227015996283e-05,
+      "loss": 1.2699,
+      "step": 1814
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8366852142565352e-05,
+      "loss": 1.2821,
+      "step": 1815
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8354473385202772e-05,
+      "loss": 1.2856,
+      "step": 1816
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.834209074965482e-05,
+      "loss": 1.3096,
+      "step": 1817
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8329704241669574e-05,
+      "loss": 1.2801,
+      "step": 1818
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8317313866996897e-05,
+      "loss": 1.2775,
+      "step": 1819
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.830491963138848e-05,
+      "loss": 1.2593,
+      "step": 1820
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8292521540597767e-05,
+      "loss": 1.2412,
+      "step": 1821
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.828011960038002e-05,
+      "loss": 1.2641,
+      "step": 1822
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.826771381649227e-05,
+      "loss": 1.2691,
+      "step": 1823
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8255304194693343e-05,
+      "loss": 1.2985,
+      "step": 1824
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8242890740743844e-05,
+      "loss": 1.2997,
+      "step": 1825
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8230473460406154e-05,
+      "loss": 1.3265,
+      "step": 1826
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8218052359444434e-05,
+      "loss": 1.3146,
+      "step": 1827
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8205627443624616e-05,
+      "loss": 1.3128,
+      "step": 1828
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8193198718714402e-05,
+      "loss": 1.2767,
+      "step": 1829
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8180766190483263e-05,
+      "loss": 1.2702,
+      "step": 1830
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8168329864702443e-05,
+      "loss": 1.3528,
+      "step": 1831
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8155889747144933e-05,
+      "loss": 1.2752,
+      "step": 1832
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8143445843585498e-05,
+      "loss": 1.2833,
+      "step": 1833
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8130998159800663e-05,
+      "loss": 1.3128,
+      "step": 1834
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8118546701568687e-05,
+      "loss": 1.2616,
+      "step": 1835
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.81060914746696e-05,
+      "loss": 1.2878,
+      "step": 1836
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8093632484885182e-05,
+      "loss": 1.2238,
+      "step": 1837
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8081169737998956e-05,
+      "loss": 1.3228,
+      "step": 1838
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8068703239796175e-05,
+      "loss": 1.242,
+      "step": 1839
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.805623299606385e-05,
+      "loss": 1.2731,
+      "step": 1840
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8043759012590723e-05,
+      "loss": 1.325,
+      "step": 1841
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.803128129516729e-05,
+      "loss": 1.275,
+      "step": 1842
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.801879984958575e-05,
+      "loss": 1.2973,
+      "step": 1843
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.800631468164005e-05,
+      "loss": 1.3062,
+      "step": 1844
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7993825797125866e-05,
+      "loss": 1.2365,
+      "step": 1845
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7981333201840595e-05,
+      "loss": 1.269,
+      "step": 1846
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7968836901583364e-05,
+      "loss": 1.2851,
+      "step": 1847
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7956336902155003e-05,
+      "loss": 1.2497,
+      "step": 1848
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7943833209358076e-05,
+      "loss": 1.2965,
+      "step": 1849
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.793132582899686e-05,
+      "loss": 1.317,
+      "step": 1850
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.791881476687733e-05,
+      "loss": 1.2614,
+      "step": 1851
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.790630002880718e-05,
+      "loss": 1.2413,
+      "step": 1852
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7893781620595818e-05,
+      "loss": 1.2893,
+      "step": 1853
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.788125954805434e-05,
+      "loss": 1.2954,
+      "step": 1854
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7868733816995553e-05,
+      "loss": 1.3131,
+      "step": 1855
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7856204433233954e-05,
+      "loss": 1.2613,
+      "step": 1856
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7843671402585747e-05,
+      "loss": 1.2626,
+      "step": 1857
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.783113473086882e-05,
+      "loss": 1.2773,
+      "step": 1858
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.781859442390276e-05,
+      "loss": 1.2473,
+      "step": 1859
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.780605048750883e-05,
+      "loss": 1.3302,
+      "step": 1860
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7793502927509988e-05,
+      "loss": 1.2935,
+      "step": 1861
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7780951749730864e-05,
+      "loss": 1.3324,
+      "step": 1862
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7768396959997783e-05,
+      "loss": 1.3192,
+      "step": 1863
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7755838564138722e-05,
+      "loss": 1.2569,
+      "step": 1864
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7743276567983354e-05,
+      "loss": 1.247,
+      "step": 1865
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7730710977363023e-05,
+      "loss": 1.3006,
+      "step": 1866
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.771814179811073e-05,
+      "loss": 1.284,
+      "step": 1867
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7705569036061137e-05,
+      "loss": 1.2943,
+      "step": 1868
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7692992697050587e-05,
+      "loss": 1.3039,
+      "step": 1869
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7680412786917074e-05,
+      "loss": 1.2616,
+      "step": 1870
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7667829311500255e-05,
+      "loss": 1.2808,
+      "step": 1871
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.765524227664143e-05,
+      "loss": 1.2543,
+      "step": 1872
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7642651688183558e-05,
+      "loss": 1.2762,
+      "step": 1873
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.763005755197126e-05,
+      "loss": 1.2933,
+      "step": 1874
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7617459873850792e-05,
+      "loss": 1.3022,
+      "step": 1875
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.760485865967004e-05,
+      "loss": 1.2794,
+      "step": 1876
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7592253915278556e-05,
+      "loss": 1.2269,
+      "step": 1877
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7579645646527522e-05,
+      "loss": 1.2871,
+      "step": 1878
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7567033859269754e-05,
+      "loss": 1.2604,
+      "step": 1879
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.75544185593597e-05,
+      "loss": 1.2772,
+      "step": 1880
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.754179975265344e-05,
+      "loss": 1.2776,
+      "step": 1881
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.752917744500868e-05,
+      "loss": 1.2773,
+      "step": 1882
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7516551642284765e-05,
+      "loss": 1.2553,
+      "step": 1883
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7503922350342645e-05,
+      "loss": 1.2927,
+      "step": 1884
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7491289575044893e-05,
+      "loss": 1.2888,
+      "step": 1885
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7478653322255707e-05,
+      "loss": 1.2425,
+      "step": 1886
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.746601359784089e-05,
+      "loss": 1.2898,
+      "step": 1887
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.745337040766787e-05,
+      "loss": 1.2953,
+      "step": 1888
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.744072375760566e-05,
+      "loss": 1.3026,
+      "step": 1889
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.74280736535249e-05,
+      "loss": 1.3024,
+      "step": 1890
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7415420101297836e-05,
+      "loss": 1.2661,
+      "step": 1891
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7402763106798295e-05,
+      "loss": 1.3313,
+      "step": 1892
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.739010267590171e-05,
+      "loss": 1.323,
+      "step": 1893
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7377438814485117e-05,
+      "loss": 1.2621,
+      "step": 1894
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7364771528427145e-05,
+      "loss": 1.2979,
+      "step": 1895
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7352100823608006e-05,
+      "loss": 1.2774,
+      "step": 1896
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.733942670590949e-05,
+      "loss": 1.3335,
+      "step": 1897
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7326749181214992e-05,
+      "loss": 1.2822,
+      "step": 1898
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7314068255409466e-05,
+      "loss": 1.2991,
+      "step": 1899
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7301383934379475e-05,
+      "loss": 1.2451,
+      "step": 1900
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7288696224013124e-05,
+      "loss": 1.3066,
+      "step": 1901
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.727600513020011e-05,
+      "loss": 1.2819,
+      "step": 1902
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7263310658831697e-05,
+      "loss": 1.2558,
+      "step": 1903
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.725061281580073e-05,
+      "loss": 1.3023,
+      "step": 1904
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7237911607001586e-05,
+      "loss": 1.2866,
+      "step": 1905
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.722520703833024e-05,
+      "loss": 1.2923,
+      "step": 1906
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7212499115684204e-05,
+      "loss": 1.2528,
+      "step": 1907
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.719978784496257e-05,
+      "loss": 1.2347,
+      "step": 1908
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.718707323206595e-05,
+      "loss": 1.2529,
+      "step": 1909
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.717435528289653e-05,
+      "loss": 1.2897,
+      "step": 1910
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7161634003358056e-05,
+      "loss": 1.2463,
+      "step": 1911
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7148909399355785e-05,
+      "loss": 1.2654,
+      "step": 1912
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.713618147679655e-05,
+      "loss": 1.259,
+      "step": 1913
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.712345024158871e-05,
+      "loss": 1.2387,
+      "step": 1914
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.711071569964216e-05,
+      "loss": 1.3236,
+      "step": 1915
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7097977856868336e-05,
+      "loss": 1.3116,
+      "step": 1916
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.70852367191802e-05,
+      "loss": 1.3156,
+      "step": 1917
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.707249229249225e-05,
+      "loss": 1.2957,
+      "step": 1918
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7059744582720515e-05,
+      "loss": 1.2634,
+      "step": 1919
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7046993595782532e-05,
+      "loss": 1.2888,
+      "step": 1920
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7034239337597378e-05,
+      "loss": 1.2981,
+      "step": 1921
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7021481814085622e-05,
+      "loss": 1.1841,
+      "step": 1922
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7008721031169378e-05,
+      "loss": 1.297,
+      "step": 1923
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.699595699477226e-05,
+      "loss": 1.2748,
+      "step": 1924
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.6983189710819396e-05,
+      "loss": 1.282,
+      "step": 1925
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.697041918523741e-05,
+      "loss": 1.3256,
+      "step": 1926
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6957645423954438e-05,
+      "loss": 1.2638,
+      "step": 1927
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.694486843290013e-05,
+      "loss": 1.2963,
+      "step": 1928
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6932088218005623e-05,
+      "loss": 1.277,
+      "step": 1929
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6919304785203543e-05,
+      "loss": 1.2888,
+      "step": 1930
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6906518140428027e-05,
+      "loss": 1.2757,
+      "step": 1931
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6893728289614693e-05,
+      "loss": 1.286,
+      "step": 1932
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.688093523870065e-05,
+      "loss": 1.3027,
+      "step": 1933
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6868138993624486e-05,
+      "loss": 1.3461,
+      "step": 1934
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6855339560326284e-05,
+      "loss": 1.2842,
+      "step": 1935
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6842536944747597e-05,
+      "loss": 1.3366,
+      "step": 1936
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.682973115283146e-05,
+      "loss": 1.2429,
+      "step": 1937
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6816922190522386e-05,
+      "loss": 1.2845,
+      "step": 1938
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6804110063766345e-05,
+      "loss": 1.3069,
+      "step": 1939
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.679129477851079e-05,
+      "loss": 1.2499,
+      "step": 1940
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6778476340704636e-05,
+      "loss": 1.2346,
+      "step": 1941
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6765654756298264e-05,
+      "loss": 1.2508,
+      "step": 1942
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.675283003124351e-05,
+      "loss": 1.2743,
+      "step": 1943
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6740002171493676e-05,
+      "loss": 1.274,
+      "step": 1944
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6727171183003502e-05,
+      "loss": 1.2814,
+      "step": 1945
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6714337071729207e-05,
+      "loss": 1.3031,
+      "step": 1946
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6701499843628443e-05,
+      "loss": 1.2805,
+      "step": 1947
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.66886595046603e-05,
+      "loss": 1.2729,
+      "step": 1948
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6675816060785327e-05,
+      "loss": 1.312,
+      "step": 1949
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.666296951796552e-05,
+      "loss": 1.2742,
+      "step": 1950
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6650119882164292e-05,
+      "loss": 1.292,
+      "step": 1951
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.663726715934651e-05,
+      "loss": 1.2969,
+      "step": 1952
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6624411355478463e-05,
+      "loss": 1.2734,
+      "step": 1953
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.661155247652788e-05,
+      "loss": 1.293,
+      "step": 1954
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6598690528463916e-05,
+      "loss": 1.2688,
+      "step": 1955
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6585825517257133e-05,
+      "loss": 1.2413,
+      "step": 1956
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6572957448879547e-05,
+      "loss": 1.2887,
+      "step": 1957
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.656008632930456e-05,
+      "loss": 1.318,
+      "step": 1958
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.654721216450701e-05,
+      "loss": 1.267,
+      "step": 1959
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.653433496046315e-05,
+      "loss": 1.265,
+      "step": 1960
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.652145472315063e-05,
+      "loss": 1.2791,
+      "step": 1961
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.650857145854852e-05,
+      "loss": 1.2736,
+      "step": 1962
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6495685172637292e-05,
+      "loss": 1.3451,
+      "step": 1963
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6482795871398815e-05,
+      "loss": 1.279,
+      "step": 1964
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.646990356081637e-05,
+      "loss": 1.2973,
+      "step": 1965
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.645700824687462e-05,
+      "loss": 1.278,
+      "step": 1966
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.644410993555963e-05,
+      "loss": 1.2729,
+      "step": 1967
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.643120863285886e-05,
+      "loss": 1.3152,
+      "step": 1968
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6418304344761165e-05,
+      "loss": 1.3012,
+      "step": 1969
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6405397077256752e-05,
+      "loss": 1.2313,
+      "step": 1970
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6392486836337256e-05,
+      "loss": 1.2733,
+      "step": 1971
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.637957362799566e-05,
+      "loss": 1.2687,
+      "step": 1972
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.636665745822633e-05,
+      "loss": 1.3039,
+      "step": 1973
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6353738333025022e-05,
+      "loss": 1.2624,
+      "step": 1974
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6340816258388858e-05,
+      "loss": 1.3124,
+      "step": 1975
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6327891240316313e-05,
+      "loss": 1.2835,
+      "step": 1976
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6314963284807246e-05,
+      "loss": 1.2391,
+      "step": 1977
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.630203239786287e-05,
+      "loss": 1.2639,
+      "step": 1978
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.628909858548577e-05,
+      "loss": 1.285,
+      "step": 1979
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6276161853679877e-05,
+      "loss": 1.2698,
+      "step": 1980
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.626322220845048e-05,
+      "loss": 1.2697,
+      "step": 1981
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6250279655804232e-05,
+      "loss": 1.3049,
+      "step": 1982
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6237334201749126e-05,
+      "loss": 1.3179,
+      "step": 1983
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6224385852294484e-05,
+      "loss": 1.3303,
+      "step": 1984
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6211434613451006e-05,
+      "loss": 1.2867,
+      "step": 1985
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6198480491230712e-05,
+      "loss": 1.2793,
+      "step": 1986
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.618552349164697e-05,
+      "loss": 1.2727,
+      "step": 1987
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6172563620714475e-05,
+      "loss": 1.2906,
+      "step": 1988
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6159600884449258e-05,
+      "loss": 1.266,
+      "step": 1989
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6146635288868685e-05,
+      "loss": 1.2422,
+      "step": 1990
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6133666839991444e-05,
+      "loss": 1.2764,
+      "step": 1991
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.612069554383755e-05,
+      "loss": 1.2452,
+      "step": 1992
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6107721406428338e-05,
+      "loss": 1.3092,
+      "step": 1993
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6094744433786467e-05,
+      "loss": 1.2806,
+      "step": 1994
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6081764631935896e-05,
+      "loss": 1.2924,
+      "step": 1995
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.606878200690193e-05,
+      "loss": 1.2903,
+      "step": 1996
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.605579656471115e-05,
+      "loss": 1.2397,
+      "step": 1997
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6042808311391456e-05,
+      "loss": 1.2608,
+      "step": 1998
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6029817252972064e-05,
+      "loss": 1.2555,
+      "step": 1999
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6016823395483482e-05,
+      "loss": 1.2567,
+      "step": 2000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.600382674495751e-05,
+      "loss": 1.3133,
+      "step": 2001
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5990827307427263e-05,
+      "loss": 1.3379,
+      "step": 2002
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5977825088927135e-05,
+      "loss": 1.1998,
+      "step": 2003
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5964820095492825e-05,
+      "loss": 1.276,
+      "step": 2004
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5951812333161298e-05,
+      "loss": 1.2713,
+      "step": 2005
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.593880180797083e-05,
+      "loss": 1.2559,
+      "step": 2006
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5925788525960964e-05,
+      "loss": 1.3017,
+      "step": 2007
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5912772493172523e-05,
+      "loss": 1.3025,
+      "step": 2008
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5899753715647614e-05,
+      "loss": 1.3044,
+      "step": 2009
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5886732199429606e-05,
+      "loss": 1.2433,
+      "step": 2010
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.587370795056315e-05,
+      "loss": 1.2535,
+      "step": 2011
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5860680975094178e-05,
+      "loss": 1.2794,
+      "step": 2012
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5847651279069847e-05,
+      "loss": 1.2642,
+      "step": 2013
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5834618868538623e-05,
+      "loss": 1.3007,
+      "step": 2014
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.58215837495502e-05,
+      "loss": 1.2566,
+      "step": 2015
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5808545928155547e-05,
+      "loss": 1.3239,
+      "step": 2016
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5795505410406878e-05,
+      "loss": 1.2546,
+      "step": 2017
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5782462202357664e-05,
+      "loss": 1.2982,
+      "step": 2018
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5769416310062622e-05,
+      "loss": 1.3559,
+      "step": 2019
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5756367739577713e-05,
+      "loss": 1.259,
+      "step": 2020
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5743316496960154e-05,
+      "loss": 1.3018,
+      "step": 2021
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.573026258826838e-05,
+      "loss": 1.2526,
+      "step": 2022
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.571720601956208e-05,
+      "loss": 1.2777,
+      "step": 2023
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.570414679690218e-05,
+      "loss": 1.2883,
+      "step": 2024
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5691084926350825e-05,
+      "loss": 1.2386,
+      "step": 2025
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.56780204139714e-05,
+      "loss": 1.2968,
+      "step": 2026
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5664953265828504e-05,
+      "loss": 1.2461,
+      "step": 2027
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.565188348798798e-05,
+      "loss": 1.2403,
+      "step": 2028
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5638811086516873e-05,
+      "loss": 1.2555,
+      "step": 2029
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.562573606748345e-05,
+      "loss": 1.2388,
+      "step": 2030
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5612658436957204e-05,
+      "loss": 1.2985,
+      "step": 2031
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5599578201008824e-05,
+      "loss": 1.3127,
+      "step": 2032
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5586495365710225e-05,
+      "loss": 1.297,
+      "step": 2033
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5573409937134508e-05,
+      "loss": 1.2838,
+      "step": 2034
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5560321921355996e-05,
+      "loss": 1.2908,
+      "step": 2035
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.554723132445021e-05,
+      "loss": 1.2932,
+      "step": 2036
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5534138152493863e-05,
+      "loss": 1.2327,
+      "step": 2037
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5521042411564866e-05,
+      "loss": 1.2842,
+      "step": 2038
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5507944107742314e-05,
+      "loss": 1.2342,
+      "step": 2039
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.549484324710652e-05,
+      "loss": 1.3055,
+      "step": 2040
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.548173983573895e-05,
+      "loss": 1.2898,
+      "step": 2041
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5468633879722272e-05,
+      "loss": 1.2622,
+      "step": 2042
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.545552538514033e-05,
+      "loss": 1.3072,
+      "step": 2043
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5442414358078148e-05,
+      "loss": 1.2473,
+      "step": 2044
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5429300804621934e-05,
+      "loss": 1.2588,
+      "step": 2045
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.541618473085905e-05,
+      "loss": 1.293,
+      "step": 2046
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5403066142878047e-05,
+      "loss": 1.2812,
+      "step": 2047
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.538994504676862e-05,
+      "loss": 1.2819,
+      "step": 2048
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.537682144862166e-05,
+      "loss": 1.2955,
+      "step": 2049
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.53636953545292e-05,
+      "loss": 1.2705,
+      "step": 2050
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5350566770584423e-05,
+      "loss": 1.303,
+      "step": 2051
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5337435702881683e-05,
+      "loss": 1.2637,
+      "step": 2052
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5324302157516486e-05,
+      "loss": 1.2481,
+      "step": 2053
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.531116614058548e-05,
+      "loss": 1.3105,
+      "step": 2054
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5298027658186472e-05,
+      "loss": 1.2549,
+      "step": 2055
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.52848867164184e-05,
+      "loss": 1.2724,
+      "step": 2056
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5271743321381354e-05,
+      "loss": 1.2496,
+      "step": 2057
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.525859747917656e-05,
+      "loss": 1.2614,
+      "step": 2058
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.524544919590638e-05,
+      "loss": 1.2755,
+      "step": 2059
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5232298477674297e-05,
+      "loss": 1.2589,
+      "step": 2060
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5219145330584945e-05,
+      "loss": 1.2813,
+      "step": 2061
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5205989760744084e-05,
+      "loss": 1.2259,
+      "step": 2062
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5192831774258575e-05,
+      "loss": 1.2194,
+      "step": 2063
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5179671377236422e-05,
+      "loss": 1.2996,
+      "step": 2064
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.516650857578674e-05,
+      "loss": 1.2554,
+      "step": 2065
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.515334337601977e-05,
+      "loss": 1.2868,
+      "step": 2066
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5140175784046858e-05,
+      "loss": 1.2312,
+      "step": 2067
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.512700580598045e-05,
+      "loss": 1.3,
+      "step": 2068
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.5113833447934126e-05,
+      "loss": 1.3146,
+      "step": 2069
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.510065871602255e-05,
+      "loss": 1.2799,
+      "step": 2070
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.5087481616361493e-05,
+      "loss": 1.2765,
+      "step": 2071
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.5074302155067823e-05,
+      "loss": 1.2673,
+      "step": 2072
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.5061120338259512e-05,
+      "loss": 1.2831,
+      "step": 2073
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.5047936172055613e-05,
+      "loss": 1.2831,
+      "step": 2074
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.5034749662576293e-05,
+      "loss": 1.3215,
+      "step": 2075
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.5021560815942777e-05,
+      "loss": 1.2723,
+      "step": 2076
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.5008369638277382e-05,
+      "loss": 1.2905,
+      "step": 2077
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4995176135703533e-05,
+      "loss": 1.2612,
+      "step": 2078
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.498198031434571e-05,
+      "loss": 1.3296,
+      "step": 2079
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.496878218032947e-05,
+      "loss": 1.2701,
+      "step": 2080
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.495558173978145e-05,
+      "loss": 1.2911,
+      "step": 2081
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.494237899882935e-05,
+      "loss": 1.3086,
+      "step": 2082
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4929173963601958e-05,
+      "loss": 1.2568,
+      "step": 2083
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4915966640229098e-05,
+      "loss": 1.2638,
+      "step": 2084
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4902757034841674e-05,
+      "loss": 1.2551,
+      "step": 2085
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4889545153571657e-05,
+      "loss": 1.3159,
+      "step": 2086
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4876331002552055e-05,
+      "loss": 1.2429,
+      "step": 2087
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4863114587916933e-05,
+      "loss": 1.2593,
+      "step": 2088
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.484989591580142e-05,
+      "loss": 1.29,
+      "step": 2089
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4836674992341684e-05,
+      "loss": 1.2493,
+      "step": 2090
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4823451823674943e-05,
+      "loss": 1.2658,
+      "step": 2091
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.481022641593944e-05,
+      "loss": 1.2678,
+      "step": 2092
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4796998775274482e-05,
+      "loss": 1.2909,
+      "step": 2093
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4783768907820403e-05,
+      "loss": 1.3031,
+      "step": 2094
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4770536819718562e-05,
+      "loss": 1.2859,
+      "step": 2095
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.475730251711136e-05,
+      "loss": 1.2783,
+      "step": 2096
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4744066006142218e-05,
+      "loss": 1.2627,
+      "step": 2097
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4730827292955592e-05,
+      "loss": 1.2579,
+      "step": 2098
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4717586383696947e-05,
+      "loss": 1.2372,
+      "step": 2099
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.470434328451278e-05,
+      "loss": 1.2901,
+      "step": 2100
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4691098001550588e-05,
+      "loss": 1.2395,
+      "step": 2101
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4677850540958906e-05,
+      "loss": 1.2388,
+      "step": 2102
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4664600908887272e-05,
+      "loss": 1.2986,
+      "step": 2103
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4651349111486212e-05,
+      "loss": 1.3011,
+      "step": 2104
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4638095154907276e-05,
+      "loss": 1.3088,
+      "step": 2105
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4624839045303014e-05,
+      "loss": 1.3081,
+      "step": 2106
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4611580788826973e-05,
+      "loss": 1.2722,
+      "step": 2107
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4598320391633702e-05,
+      "loss": 1.2768,
+      "step": 2108
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4585057859878732e-05,
+      "loss": 1.3224,
+      "step": 2109
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4571793199718593e-05,
+      "loss": 1.3039,
+      "step": 2110
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4558526417310805e-05,
+      "loss": 1.2917,
+      "step": 2111
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4545257518813866e-05,
+      "loss": 1.2374,
+      "step": 2112
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4531986510387268e-05,
+      "loss": 1.2449,
+      "step": 2113
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4518713398191464e-05,
+      "loss": 1.2528,
+      "step": 2114
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.45054381883879e-05,
+      "loss": 1.2509,
+      "step": 2115
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4492160887138998e-05,
+      "loss": 1.2794,
+      "step": 2116
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.447888150060813e-05,
+      "loss": 1.2612,
+      "step": 2117
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4465600034959654e-05,
+      "loss": 1.2718,
+      "step": 2118
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4452316496358885e-05,
+      "loss": 1.313,
+      "step": 2119
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.443903089097211e-05,
+      "loss": 1.2596,
+      "step": 2120
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4425743224966567e-05,
+      "loss": 1.267,
+      "step": 2121
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4412453504510447e-05,
+      "loss": 1.2455,
+      "step": 2122
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.43991617357729e-05,
+      "loss": 1.3376,
+      "step": 2123
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4385867924924037e-05,
+      "loss": 1.3297,
+      "step": 2124
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.43725720781349e-05,
+      "loss": 1.2579,
+      "step": 2125
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4359274201577478e-05,
+      "loss": 1.3401,
+      "step": 2126
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4345974301424717e-05,
+      "loss": 1.2648,
+      "step": 2127
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.433267238385048e-05,
+      "loss": 1.2432,
+      "step": 2128
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4319368455029598e-05,
+      "loss": 1.2588,
+      "step": 2129
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.43060625211378e-05,
+      "loss": 1.288,
+      "step": 2130
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4292754588351768e-05,
+      "loss": 1.2282,
+      "step": 2131
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.42794446628491e-05,
+      "loss": 1.293,
+      "step": 2132
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.426613275080834e-05,
+      "loss": 1.307,
+      "step": 2133
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4252818858408923e-05,
+      "loss": 1.3079,
+      "step": 2134
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4239502991831233e-05,
+      "loss": 1.284,
+      "step": 2135
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4226185157256546e-05,
+      "loss": 1.2037,
+      "step": 2136
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.421286536086707e-05,
+      "loss": 1.2519,
+      "step": 2137
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4199543608845916e-05,
+      "loss": 1.2597,
+      "step": 2138
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4186219907377097e-05,
+      "loss": 1.2732,
+      "step": 2139
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4172894262645544e-05,
+      "loss": 1.3214,
+      "step": 2140
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4159566680837086e-05,
+      "loss": 1.278,
+      "step": 2141
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.414623716813844e-05,
+      "loss": 1.2699,
+      "step": 2142
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.413290573073723e-05,
+      "loss": 1.3015,
+      "step": 2143
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4119572374821968e-05,
+      "loss": 1.2712,
+      "step": 2144
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4106237106582072e-05,
+      "loss": 1.2545,
+      "step": 2145
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4092899932207824e-05,
+      "loss": 1.2726,
+      "step": 2146
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4079560857890405e-05,
+      "loss": 1.2493,
+      "step": 2147
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.406621988982188e-05,
+      "loss": 1.2737,
+      "step": 2148
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.405287703419518e-05,
+      "loss": 1.2957,
+      "step": 2149
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4039532297204125e-05,
+      "loss": 1.2383,
+      "step": 2150
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4026185685043405e-05,
+      "loss": 1.2334,
+      "step": 2151
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4012837203908582e-05,
+      "loss": 1.2657,
+      "step": 2152
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3999486859996073e-05,
+      "loss": 1.2939,
+      "step": 2153
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3986134659503187e-05,
+      "loss": 1.246,
+      "step": 2154
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3972780608628057e-05,
+      "loss": 1.3431,
+      "step": 2155
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3959424713569708e-05,
+      "loss": 1.2937,
+      "step": 2156
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3946066980528e-05,
+      "loss": 1.2899,
+      "step": 2157
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3932707415703673e-05,
+      "loss": 1.2751,
+      "step": 2158
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.391934602529828e-05,
+      "loss": 1.2325,
+      "step": 2159
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3905982815514243e-05,
+      "loss": 1.2628,
+      "step": 2160
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3892617792554833e-05,
+      "loss": 1.2681,
+      "step": 2161
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3879250962624152e-05,
+      "loss": 1.29,
+      "step": 2162
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.386588233192715e-05,
+      "loss": 1.2623,
+      "step": 2163
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.38525119066696e-05,
+      "loss": 1.29,
+      "step": 2164
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3839139693058116e-05,
+      "loss": 1.268,
+      "step": 2165
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.382576569730015e-05,
+      "loss": 1.2672,
+      "step": 2166
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3812389925603963e-05,
+      "loss": 1.226,
+      "step": 2167
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3799012384178654e-05,
+      "loss": 1.2533,
+      "step": 2168
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3785633079234144e-05,
+      "loss": 1.2878,
+      "step": 2169
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.377225201698117e-05,
+      "loss": 1.2776,
+      "step": 2170
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3758869203631266e-05,
+      "loss": 1.3176,
+      "step": 2171
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3745484645396816e-05,
+      "loss": 1.2473,
+      "step": 2172
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.373209834849098e-05,
+      "loss": 1.2396,
+      "step": 2173
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3718710319127755e-05,
+      "loss": 1.2944,
+      "step": 2174
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.370532056352191e-05,
+      "loss": 1.2887,
+      "step": 2175
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3691929087889042e-05,
+      "loss": 1.2868,
+      "step": 2176
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3678535898445533e-05,
+      "loss": 1.2609,
+      "step": 2177
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3665141001408562e-05,
+      "loss": 1.2767,
+      "step": 2178
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3651744402996114e-05,
+      "loss": 1.2674,
+      "step": 2179
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3638346109426932e-05,
+      "loss": 1.2839,
+      "step": 2180
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.362494612692058e-05,
+      "loss": 1.3094,
+      "step": 2181
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.361154446169739e-05,
+      "loss": 1.2948,
+      "step": 2182
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3598141119978482e-05,
+      "loss": 1.3063,
+      "step": 2183
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3584736107985737e-05,
+      "loss": 1.2767,
+      "step": 2184
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.357132943194183e-05,
+      "loss": 1.2409,
+      "step": 2185
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.35579210980702e-05,
+      "loss": 1.3158,
+      "step": 2186
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3544511112595068e-05,
+      "loss": 1.2893,
+      "step": 2187
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3531099481741403e-05,
+      "loss": 1.2235,
+      "step": 2188
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.351768621173495e-05,
+      "loss": 1.2892,
+      "step": 2189
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3504271308802204e-05,
+      "loss": 1.2643,
+      "step": 2190
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3490854779170436e-05,
+      "loss": 1.293,
+      "step": 2191
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.347743662906765e-05,
+      "loss": 1.2984,
+      "step": 2192
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3464016864722625e-05,
+      "loss": 1.2651,
+      "step": 2193
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.345059549236487e-05,
+      "loss": 1.2821,
+      "step": 2194
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.343717251822465e-05,
+      "loss": 1.2066,
+      "step": 2195
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3423747948532976e-05,
+      "loss": 1.2939,
+      "step": 2196
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.341032178952159e-05,
+      "loss": 1.2487,
+      "step": 2197
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.339689404742298e-05,
+      "loss": 1.2796,
+      "step": 2198
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.338346472847037e-05,
+      "loss": 1.2045,
+      "step": 2199
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3370033838897702e-05,
+      "loss": 1.2245,
+      "step": 2200
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3356601384939665e-05,
+      "loss": 1.3075,
+      "step": 2201
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3343167372831665e-05,
+      "loss": 1.309,
+      "step": 2202
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3329731808809836e-05,
+      "loss": 1.3084,
+      "step": 2203
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.331629469911103e-05,
+      "loss": 1.2633,
+      "step": 2204
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.330285604997281e-05,
+      "loss": 1.2671,
+      "step": 2205
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.328941586763346e-05,
+      "loss": 1.2656,
+      "step": 2206
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3275974158331977e-05,
+      "loss": 1.2829,
+      "step": 2207
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3262530928308068e-05,
+      "loss": 1.2457,
+      "step": 2208
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3249086183802137e-05,
+      "loss": 1.2854,
+      "step": 2209
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.32356399310553e-05,
+      "loss": 1.2401,
+      "step": 2210
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3222192176309367e-05,
+      "loss": 1.2986,
+      "step": 2211
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.320874292580685e-05,
+      "loss": 1.3068,
+      "step": 2212
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3195292185790957e-05,
+      "loss": 1.3193,
+      "step": 2213
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.318183996250558e-05,
+      "loss": 1.2133,
+      "step": 2214
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3168386262195307e-05,
+      "loss": 1.2821,
+      "step": 2215
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.315493109110541e-05,
+      "loss": 1.2341,
+      "step": 2216
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.314147445548183e-05,
+      "loss": 1.2578,
+      "step": 2217
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3128016361571213e-05,
+      "loss": 1.3141,
+      "step": 2218
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3114556815620863e-05,
+      "loss": 1.3132,
+      "step": 2219
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3101095823878764e-05,
+      "loss": 1.3046,
+      "step": 2220
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.308763339259357e-05,
+      "loss": 1.2741,
+      "step": 2221
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3074169528014605e-05,
+      "loss": 1.2767,
+      "step": 2222
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.306070423639186e-05,
+      "loss": 1.264,
+      "step": 2223
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3047237523975984e-05,
+      "loss": 1.263,
+      "step": 2224
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3033769397018286e-05,
+      "loss": 1.2069,
+      "step": 2225
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3020299861770732e-05,
+      "loss": 1.2284,
+      "step": 2226
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.300682892448595e-05,
+      "loss": 1.2452,
+      "step": 2227
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2993356591417203e-05,
+      "loss": 1.2717,
+      "step": 2228
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2979882868818422e-05,
+      "loss": 1.263,
+      "step": 2229
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.296640776294416e-05,
+      "loss": 1.2966,
+      "step": 2230
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2952931280049628e-05,
+      "loss": 1.2989,
+      "step": 2231
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.293945342639067e-05,
+      "loss": 1.2976,
+      "step": 2232
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2925974208223778e-05,
+      "loss": 1.27,
+      "step": 2233
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2912493631806055e-05,
+      "loss": 1.2463,
+      "step": 2234
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2899011703395254e-05,
+      "loss": 1.2726,
+      "step": 2235
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.288552842924974e-05,
+      "loss": 1.2412,
+      "step": 2236
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2872043815628525e-05,
+      "loss": 1.2547,
+      "step": 2237
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2858557868791222e-05,
+      "loss": 1.289,
+      "step": 2238
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.284507059499807e-05,
+      "loss": 1.2979,
+      "step": 2239
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.283158200050993e-05,
+      "loss": 1.2521,
+      "step": 2240
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2818092091588266e-05,
+      "loss": 1.2675,
+      "step": 2241
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.280460087449515e-05,
+      "loss": 1.2783,
+      "step": 2242
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2791108355493278e-05,
+      "loss": 1.2532,
+      "step": 2243
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2777614540845934e-05,
+      "loss": 1.2963,
+      "step": 2244
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2764119436817015e-05,
+      "loss": 1.2672,
+      "step": 2245
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2750623049671003e-05,
+      "loss": 1.2974,
+      "step": 2246
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.273712538567299e-05,
+      "loss": 1.2806,
+      "step": 2247
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2723626451088644e-05,
+      "loss": 1.3214,
+      "step": 2248
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2710126252184255e-05,
+      "loss": 1.307,
+      "step": 2249
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2696624795226662e-05,
+      "loss": 1.2484,
+      "step": 2250
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2683122086483297e-05,
+      "loss": 1.24,
+      "step": 2251
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.26696181322222e-05,
+      "loss": 1.2795,
+      "step": 2252
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2656112938711952e-05,
+      "loss": 1.2912,
+      "step": 2253
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.264260651222174e-05,
+      "loss": 1.2853,
+      "step": 2254
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.26290988590213e-05,
+      "loss": 1.2958,
+      "step": 2255
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.261558998538095e-05,
+      "loss": 1.2503,
+      "step": 2256
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2602079897571576e-05,
+      "loss": 1.295,
+      "step": 2257
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.258856860186462e-05,
+      "loss": 1.269,
+      "step": 2258
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.257505610453209e-05,
+      "loss": 1.2443,
+      "step": 2259
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2561542411846537e-05,
+      "loss": 1.3001,
+      "step": 2260
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.25480275300811e-05,
+      "loss": 1.2766,
+      "step": 2261
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.253451146550945e-05,
+      "loss": 1.3115,
+      "step": 2262
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.252099422440579e-05,
+      "loss": 1.3036,
+      "step": 2263
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2507475813044896e-05,
+      "loss": 1.3008,
+      "step": 2264
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2493956237702075e-05,
+      "loss": 1.2758,
+      "step": 2265
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2480435504653185e-05,
+      "loss": 1.2717,
+      "step": 2266
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.24669136201746e-05,
+      "loss": 1.2827,
+      "step": 2267
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2453390590543246e-05,
+      "loss": 1.2494,
+      "step": 2268
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.243986642203658e-05,
+      "loss": 1.2719,
+      "step": 2269
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2426341120932582e-05,
+      "loss": 1.2765,
+      "step": 2270
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.241281469350976e-05,
+      "loss": 1.2572,
+      "step": 2271
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2399287146047137e-05,
+      "loss": 1.2562,
+      "step": 2272
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2385758484824275e-05,
+      "loss": 1.2728,
+      "step": 2273
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2372228716121246e-05,
+      "loss": 1.3079,
+      "step": 2274
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.235869784621861e-05,
+      "loss": 1.2781,
+      "step": 2275
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2345165881397475e-05,
+      "loss": 1.298,
+      "step": 2276
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2331632827939438e-05,
+      "loss": 1.2634,
+      "step": 2277
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.231809869212661e-05,
+      "loss": 1.2548,
+      "step": 2278
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.230456348024159e-05,
+      "loss": 1.2609,
+      "step": 2279
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2291027198567502e-05,
+      "loss": 1.304,
+      "step": 2280
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2277489853387932e-05,
+      "loss": 1.2894,
+      "step": 2281
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2263951450986987e-05,
+      "loss": 1.2655,
+      "step": 2282
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2250411997649266e-05,
+      "loss": 1.2448,
+      "step": 2283
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2236871499659824e-05,
+      "loss": 1.3025,
+      "step": 2284
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2223329963304242e-05,
+      "loss": 1.2805,
+      "step": 2285
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2209787394868562e-05,
+      "loss": 1.2748,
+      "step": 2286
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2196243800639303e-05,
+      "loss": 1.2832,
+      "step": 2287
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2182699186903462e-05,
+      "loss": 1.2327,
+      "step": 2288
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2169153559948513e-05,
+      "loss": 1.303,
+      "step": 2289
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.21556069260624e-05,
+      "loss": 1.2921,
+      "step": 2290
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2142059291533542e-05,
+      "loss": 1.2519,
+      "step": 2291
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2128510662650796e-05,
+      "loss": 1.2517,
+      "step": 2292
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.211496104570351e-05,
+      "loss": 1.2842,
+      "step": 2293
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.210141044698148e-05,
+      "loss": 1.2987,
+      "step": 2294
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2087858872774954e-05,
+      "loss": 1.2548,
+      "step": 2295
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2074306329374636e-05,
+      "loss": 1.3158,
+      "step": 2296
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.206075282307168e-05,
+      "loss": 1.2867,
+      "step": 2297
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2047198360157683e-05,
+      "loss": 1.282,
+      "step": 2298
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2033642946924698e-05,
+      "loss": 1.2701,
+      "step": 2299
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2020086589665203e-05,
+      "loss": 1.23,
+      "step": 2300
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2006529294672126e-05,
+      "loss": 1.2628,
+      "step": 2301
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.1992971068238826e-05,
+      "loss": 1.2468,
+      "step": 2302
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.197941191665909e-05,
+      "loss": 1.269,
+      "step": 2303
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.196585184622715e-05,
+      "loss": 1.2319,
+      "step": 2304
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.195229086323764e-05,
+      "loss": 1.2721,
+      "step": 2305
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.193872897398564e-05,
+      "loss": 1.2655,
+      "step": 2306
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1925166184766636e-05,
+      "loss": 1.2863,
+      "step": 2307
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1911602501876546e-05,
+      "loss": 1.2351,
+      "step": 2308
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1898037931611688e-05,
+      "loss": 1.2753,
+      "step": 2309
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1884472480268806e-05,
+      "loss": 1.2352,
+      "step": 2310
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1870906154145035e-05,
+      "loss": 1.2392,
+      "step": 2311
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.185733895953794e-05,
+      "loss": 1.2358,
+      "step": 2312
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1843770902745462e-05,
+      "loss": 1.283,
+      "step": 2313
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1830201990065966e-05,
+      "loss": 1.2299,
+      "step": 2314
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1816632227798196e-05,
+      "loss": 1.3021,
+      "step": 2315
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.180306162224131e-05,
+      "loss": 1.2538,
+      "step": 2316
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1789490179694833e-05,
+      "loss": 1.2436,
+      "step": 2317
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1775917906458698e-05,
+      "loss": 1.2869,
+      "step": 2318
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.176234480883322e-05,
+      "loss": 1.2631,
+      "step": 2319
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.174877089311909e-05,
+      "loss": 1.2608,
+      "step": 2320
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1735196165617385e-05,
+      "loss": 1.1975,
+      "step": 2321
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1721620632629552e-05,
+      "loss": 1.2808,
+      "step": 2322
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1708044300457423e-05,
+      "loss": 1.2342,
+      "step": 2323
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1694467175403197e-05,
+      "loss": 1.3237,
+      "step": 2324
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1680889263769425e-05,
+      "loss": 1.2427,
+      "step": 2325
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.166731057185905e-05,
+      "loss": 1.2862,
+      "step": 2326
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1653731105975355e-05,
+      "loss": 1.3005,
+      "step": 2327
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1640150872421997e-05,
+      "loss": 1.2493,
+      "step": 2328
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1626569877502985e-05,
+      "loss": 1.2957,
+      "step": 2329
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.161298812752267e-05,
+      "loss": 1.2392,
+      "step": 2330
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1599405628785773e-05,
+      "loss": 1.247,
+      "step": 2331
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.158582238759735e-05,
+      "loss": 1.3058,
+      "step": 2332
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.15722384102628e-05,
+      "loss": 1.281,
+      "step": 2333
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1558653703087876e-05,
+      "loss": 1.226,
+      "step": 2334
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1545068272378664e-05,
+      "loss": 1.308,
+      "step": 2335
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1531482124441574e-05,
+      "loss": 1.264,
+      "step": 2336
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.151789526558337e-05,
+      "loss": 1.2489,
+      "step": 2337
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1504307702111125e-05,
+      "loss": 1.2956,
+      "step": 2338
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1490719440332252e-05,
+      "loss": 1.2503,
+      "step": 2339
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.147713048655449e-05,
+      "loss": 1.2541,
+      "step": 2340
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1463540847085892e-05,
+      "loss": 1.3193,
+      "step": 2341
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1449950528234828e-05,
+      "loss": 1.3139,
+      "step": 2342
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.143635953630999e-05,
+      "loss": 1.2783,
+      "step": 2343
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1422767877620382e-05,
+      "loss": 1.2414,
+      "step": 2344
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1409175558475307e-05,
+      "loss": 1.2978,
+      "step": 2345
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1395582585184397e-05,
+      "loss": 1.2643,
+      "step": 2346
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.138198896405756e-05,
+      "loss": 1.2579,
+      "step": 2347
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1368394701405023e-05,
+      "loss": 1.2258,
+      "step": 2348
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1354799803537312e-05,
+      "loss": 1.2374,
+      "step": 2349
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.134120427676523e-05,
+      "loss": 1.2371,
+      "step": 2350
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1327608127399895e-05,
+      "loss": 1.2696,
+      "step": 2351
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1314011361752687e-05,
+      "loss": 1.2591,
+      "step": 2352
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1300413986135313e-05,
+      "loss": 1.2424,
+      "step": 2353
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.128681600685971e-05,
+      "loss": 1.3086,
+      "step": 2354
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1273217430238146e-05,
+      "loss": 1.3011,
+      "step": 2355
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1259618262583122e-05,
+      "loss": 1.2495,
+      "step": 2356
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1246018510207452e-05,
+      "loss": 1.2858,
+      "step": 2357
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1232418179424204e-05,
+      "loss": 1.2344,
+      "step": 2358
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.12188172765467e-05,
+      "loss": 1.221,
+      "step": 2359
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.120521580788856e-05,
+      "loss": 1.2757,
+      "step": 2360
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1191613779763635e-05,
+      "loss": 1.2317,
+      "step": 2361
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1178011198486064e-05,
+      "loss": 1.2473,
+      "step": 2362
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1164408070370212e-05,
+      "loss": 1.276,
+      "step": 2363
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1150804401730724e-05,
+      "loss": 1.2718,
+      "step": 2364
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1137200198882484e-05,
+      "loss": 1.2227,
+      "step": 2365
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.112359546814063e-05,
+      "loss": 1.2493,
+      "step": 2366
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.110999021582053e-05,
+      "loss": 1.276,
+      "step": 2367
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1096384448237824e-05,
+      "loss": 1.2941,
+      "step": 2368
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1082778171708355e-05,
+      "loss": 1.224,
+      "step": 2369
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1069171392548226e-05,
+      "loss": 1.2903,
+      "step": 2370
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1055564117073767e-05,
+      "loss": 1.26,
+      "step": 2371
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1041956351601543e-05,
+      "loss": 1.2548,
+      "step": 2372
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1028348102448338e-05,
+      "loss": 1.1928,
+      "step": 2373
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1014739375931166e-05,
+      "loss": 1.27,
+      "step": 2374
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1001130178367256e-05,
+      "loss": 1.2655,
+      "step": 2375
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.098752051607406e-05,
+      "loss": 1.2878,
+      "step": 2376
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.097391039536926e-05,
+      "loss": 1.3185,
+      "step": 2377
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0960299822570728e-05,
+      "loss": 1.2573,
+      "step": 2378
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.094668880399655e-05,
+      "loss": 1.294,
+      "step": 2379
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0933077345965032e-05,
+      "loss": 1.3151,
+      "step": 2380
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0919465454794672e-05,
+      "loss": 1.2751,
+      "step": 2381
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0905853136804173e-05,
+      "loss": 1.2886,
+      "step": 2382
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.089224039831244e-05,
+      "loss": 1.2623,
+      "step": 2383
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.087862724563857e-05,
+      "loss": 1.3,
+      "step": 2384
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0865013685101844e-05,
+      "loss": 1.2876,
+      "step": 2385
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.085139972302175e-05,
+      "loss": 1.3102,
+      "step": 2386
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.083778536571795e-05,
+      "loss": 1.2676,
+      "step": 2387
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0824170619510283e-05,
+      "loss": 1.2637,
+      "step": 2388
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0810555490718787e-05,
+      "loss": 1.2419,
+      "step": 2389
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0796939985663666e-05,
+      "loss": 1.2756,
+      "step": 2390
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0783324110665306e-05,
+      "loss": 1.275,
+      "step": 2391
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0769707872044242e-05,
+      "loss": 1.2722,
+      "step": 2392
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0756091276121212e-05,
+      "loss": 1.3089,
+      "step": 2393
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0742474329217094e-05,
+      "loss": 1.314,
+      "step": 2394
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0728857037652945e-05,
+      "loss": 1.2922,
+      "step": 2395
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0715239407749973e-05,
+      "loss": 1.2915,
+      "step": 2396
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.070162144582954e-05,
+      "loss": 1.2634,
+      "step": 2397
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0688003158213172e-05,
+      "loss": 1.3197,
+      "step": 2398
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.067438455122255e-05,
+      "loss": 1.2978,
+      "step": 2399
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0660765631179474e-05,
+      "loss": 1.2411,
+      "step": 2400
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0647146404405923e-05,
+      "loss": 1.2427,
+      "step": 2401
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0633526877224006e-05,
+      "loss": 1.2709,
+      "step": 2402
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.061990705595597e-05,
+      "loss": 1.2201,
+      "step": 2403
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.060628694692419e-05,
+      "loss": 1.2977,
+      "step": 2404
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0592666556451197e-05,
+      "loss": 1.2939,
+      "step": 2405
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0579045890859635e-05,
+      "loss": 1.2681,
+      "step": 2406
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0565424956472278e-05,
+      "loss": 1.3335,
+      "step": 2407
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.055180375961203e-05,
+      "loss": 1.2893,
+      "step": 2408
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.053818230660191e-05,
+      "loss": 1.2555,
+      "step": 2409
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.052456060376506e-05,
+      "loss": 1.3004,
+      "step": 2410
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.051093865742474e-05,
+      "loss": 1.2388,
+      "step": 2411
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0497316473904324e-05,
+      "loss": 1.2878,
+      "step": 2412
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.048369405952729e-05,
+      "loss": 1.2574,
+      "step": 2413
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0470071420617222e-05,
+      "loss": 1.2572,
+      "step": 2414
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.045644856349782e-05,
+      "loss": 1.2595,
+      "step": 2415
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0442825494492876e-05,
+      "loss": 1.2491,
+      "step": 2416
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0429202219926273e-05,
+      "loss": 1.2555,
+      "step": 2417
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0415578746122007e-05,
+      "loss": 1.3248,
+      "step": 2418
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0401955079404154e-05,
+      "loss": 1.2744,
+      "step": 2419
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0388331226096886e-05,
+      "loss": 1.2988,
+      "step": 2420
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0374707192524455e-05,
+      "loss": 1.2343,
+      "step": 2421
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.036108298501121e-05,
+      "loss": 1.3385,
+      "step": 2422
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.034745860988156e-05,
+      "loss": 1.2002,
+      "step": 2423
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0333834073460018e-05,
+      "loss": 1.2588,
+      "step": 2424
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.032020938207114e-05,
+      "loss": 1.2447,
+      "step": 2425
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.030658454203958e-05,
+      "loss": 1.2394,
+      "step": 2426
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.029295955969005e-05,
+      "loss": 1.2984,
+      "step": 2427
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.027933444134733e-05,
+      "loss": 1.2739,
+      "step": 2428
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0265709193336266e-05,
+      "loss": 1.3139,
+      "step": 2429
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.025208382198176e-05,
+      "loss": 1.2836,
+      "step": 2430
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0238458333608766e-05,
+      "loss": 1.3205,
+      "step": 2431
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0224832734542314e-05,
+      "loss": 1.225,
+      "step": 2432
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0211207031107457e-05,
+      "loss": 1.2714,
+      "step": 2433
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0197581229629317e-05,
+      "loss": 1.2509,
+      "step": 2434
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.018395533643305e-05,
+      "loss": 1.2631,
+      "step": 2435
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.017032935784386e-05,
+      "loss": 1.2623,
+      "step": 2436
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0156703300186997e-05,
+      "loss": 1.2872,
+      "step": 2437
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0143077169787725e-05,
+      "loss": 1.2804,
+      "step": 2438
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.012945097297137e-05,
+      "loss": 1.2716,
+      "step": 2439
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0115824716063273e-05,
+      "loss": 1.2535,
+      "step": 2440
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0102198405388806e-05,
+      "loss": 1.2782,
+      "step": 2441
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.008857204727336e-05,
+      "loss": 1.2146,
+      "step": 2442
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0074945648042353e-05,
+      "loss": 1.2816,
+      "step": 2443
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0061319214021237e-05,
+      "loss": 1.2117,
+      "step": 2444
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0047692751535454e-05,
+      "loss": 1.2778,
+      "step": 2445
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0034066266910475e-05,
+      "loss": 1.2435,
+      "step": 2446
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0020439766471775e-05,
+      "loss": 1.2687,
+      "step": 2447
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.000681325654484e-05,
+      "loss": 1.2329,
+      "step": 2448
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.999318674345516e-05,
+      "loss": 1.2704,
+      "step": 2449
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.997956023352823e-05,
+      "loss": 1.22,
+      "step": 2450
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9965933733089535e-05,
+      "loss": 1.2474,
+      "step": 2451
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.995230724846455e-05,
+      "loss": 1.1917,
+      "step": 2452
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.993868078597877e-05,
+      "loss": 1.2962,
+      "step": 2453
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9925054351957647e-05,
+      "loss": 1.289,
+      "step": 2454
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9911427952726644e-05,
+      "loss": 1.2758,
+      "step": 2455
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9897801594611204e-05,
+      "loss": 1.211,
+      "step": 2456
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.988417528393673e-05,
+      "loss": 1.2334,
+      "step": 2457
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9870549027028635e-05,
+      "loss": 1.2707,
+      "step": 2458
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9856922830212286e-05,
+      "loss": 1.2852,
+      "step": 2459
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.984329669981301e-05,
+      "loss": 1.2086,
+      "step": 2460
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9829670642156147e-05,
+      "loss": 1.2645,
+      "step": 2461
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.981604466356695e-05,
+      "loss": 1.2752,
+      "step": 2462
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.980241877037069e-05,
+      "loss": 1.1866,
+      "step": 2463
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9788792968892553e-05,
+      "loss": 1.262,
+      "step": 2464
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.977516726545769e-05,
+      "loss": 1.3017,
+      "step": 2465
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.976154166639124e-05,
+      "loss": 1.2054,
+      "step": 2466
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9747916178018246e-05,
+      "loss": 1.2443,
+      "step": 2467
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9734290806663738e-05,
+      "loss": 1.2655,
+      "step": 2468
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9720665558652676e-05,
+      "loss": 1.2588,
+      "step": 2469
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.970704044030995e-05,
+      "loss": 1.281,
+      "step": 2470
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9693415457960426e-05,
+      "loss": 1.2657,
+      "step": 2471
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9679790617928872e-05,
+      "loss": 1.2382,
+      "step": 2472
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.966616592653999e-05,
+      "loss": 1.2571,
+      "step": 2473
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9652541390118443e-05,
+      "loss": 1.2527,
+      "step": 2474
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.963891701498879e-05,
+      "loss": 1.2489,
+      "step": 2475
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9625292807475548e-05,
+      "loss": 1.2523,
+      "step": 2476
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9611668773903124e-05,
+      "loss": 1.2716,
+      "step": 2477
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9598044920595853e-05,
+      "loss": 1.203,
+      "step": 2478
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9584421253878e-05,
+      "loss": 1.3159,
+      "step": 2479
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9570797780073737e-05,
+      "loss": 1.2792,
+      "step": 2480
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.955717450550713e-05,
+      "loss": 1.2965,
+      "step": 2481
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9543551436502186e-05,
+      "loss": 1.2812,
+      "step": 2482
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9529928579382778e-05,
+      "loss": 1.2525,
+      "step": 2483
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9516305940472714e-05,
+      "loss": 1.2945,
+      "step": 2484
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9502683526095683e-05,
+      "loss": 1.3148,
+      "step": 2485
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.948906134257526e-05,
+      "loss": 1.2594,
+      "step": 2486
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.947543939623495e-05,
+      "loss": 1.2805,
+      "step": 2487
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9461817693398105e-05,
+      "loss": 1.3181,
+      "step": 2488
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.944819624038798e-05,
+      "loss": 1.2239,
+      "step": 2489
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.943457504352773e-05,
+      "loss": 1.2322,
+      "step": 2490
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.942095410914037e-05,
+      "loss": 1.2296,
+      "step": 2491
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9407333443548806e-05,
+      "loss": 1.326,
+      "step": 2492
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9393713053075816e-05,
+      "loss": 1.2577,
+      "step": 2493
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9380092944044036e-05,
+      "loss": 1.2903,
+      "step": 2494
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9366473122776e-05,
+      "loss": 1.2448,
+      "step": 2495
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9352853595594077e-05,
+      "loss": 1.2491,
+      "step": 2496
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9339234368820533e-05,
+      "loss": 1.279,
+      "step": 2497
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.932561544877746e-05,
+      "loss": 1.2312,
+      "step": 2498
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9311996841786825e-05,
+      "loss": 1.265,
+      "step": 2499
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9298378554170463e-05,
+      "loss": 1.217,
+      "step": 2500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9284760592250037e-05,
+      "loss": 1.2786,
+      "step": 2501
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9271142962347058e-05,
+      "loss": 1.273,
+      "step": 2502
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.925752567078291e-05,
+      "loss": 1.2468,
+      "step": 2503
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.924390872387879e-05,
+      "loss": 1.3012,
+      "step": 2504
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.923029212795576e-05,
+      "loss": 1.2393,
+      "step": 2505
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9216675889334704e-05,
+      "loss": 1.3048,
+      "step": 2506
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9203060014336334e-05,
+      "loss": 1.2441,
+      "step": 2507
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9189444509281216e-05,
+      "loss": 1.2865,
+      "step": 2508
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9175829380489727e-05,
+      "loss": 1.2377,
+      "step": 2509
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9162214634282055e-05,
+      "loss": 1.2666,
+      "step": 2510
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9148600276978254e-05,
+      "loss": 1.2533,
+      "step": 2511
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9134986314898156e-05,
+      "loss": 1.2998,
+      "step": 2512
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9121372754361437e-05,
+      "loss": 1.2494,
+      "step": 2513
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9107759601687562e-05,
+      "loss": 1.2354,
+      "step": 2514
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.909414686319583e-05,
+      "loss": 1.2535,
+      "step": 2515
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9080534545205334e-05,
+      "loss": 1.2787,
+      "step": 2516
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9066922654034975e-05,
+      "loss": 1.2648,
+      "step": 2517
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9053311196003457e-05,
+      "loss": 1.3116,
+      "step": 2518
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9039700177429282e-05,
+      "loss": 1.2336,
+      "step": 2519
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9026089604630743e-05,
+      "loss": 1.307,
+      "step": 2520
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9012479483925942e-05,
+      "loss": 1.2066,
+      "step": 2521
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8998869821632757e-05,
+      "loss": 1.2977,
+      "step": 2522
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.898526062406884e-05,
+      "loss": 1.2949,
+      "step": 2523
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8971651897551672e-05,
+      "loss": 1.2905,
+      "step": 2524
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8958043648398457e-05,
+      "loss": 1.2528,
+      "step": 2525
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8944435882926236e-05,
+      "loss": 1.2868,
+      "step": 2526
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.893082860745178e-05,
+      "loss": 1.1937,
+      "step": 2527
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8917221828291652e-05,
+      "loss": 1.2792,
+      "step": 2528
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8903615551762182e-05,
+      "loss": 1.2896,
+      "step": 2529
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8890009784179476e-05,
+      "loss": 1.2784,
+      "step": 2530
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8876404531859376e-05,
+      "loss": 1.3013,
+      "step": 2531
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8862799801117523e-05,
+      "loss": 1.2834,
+      "step": 2532
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.884919559826928e-05,
+      "loss": 1.2875,
+      "step": 2533
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8835591929629795e-05,
+      "loss": 1.2709,
+      "step": 2534
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.882198880151395e-05,
+      "loss": 1.227,
+      "step": 2535
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8808386220236365e-05,
+      "loss": 1.2695,
+      "step": 2536
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8794784192111448e-05,
+      "loss": 1.2871,
+      "step": 2537
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8781182723453303e-05,
+      "loss": 1.2799,
+      "step": 2538
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8767581820575803e-05,
+      "loss": 1.2406,
+      "step": 2539
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.875398148979255e-05,
+      "loss": 1.2502,
+      "step": 2540
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.874038173741688e-05,
+      "loss": 1.2613,
+      "step": 2541
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8726782569761864e-05,
+      "loss": 1.2224,
+      "step": 2542
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.87131839931403e-05,
+      "loss": 1.2702,
+      "step": 2543
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8699586013864694e-05,
+      "loss": 1.2646,
+      "step": 2544
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8685988638247316e-05,
+      "loss": 1.2868,
+      "step": 2545
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8672391872600108e-05,
+      "loss": 1.1826,
+      "step": 2546
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8658795723234774e-05,
+      "loss": 1.2546,
+      "step": 2547
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8645200196462698e-05,
+      "loss": 1.2052,
+      "step": 2548
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8631605298594977e-05,
+      "loss": 1.2719,
+      "step": 2549
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8618011035942444e-05,
+      "loss": 1.2677,
+      "step": 2550
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.860441741481561e-05,
+      "loss": 1.2691,
+      "step": 2551
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8590824441524696e-05,
+      "loss": 1.2376,
+      "step": 2552
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8577232122379625e-05,
+      "loss": 1.2478,
+      "step": 2553
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8563640463690015e-05,
+      "loss": 1.2427,
+      "step": 2554
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8550049471765176e-05,
+      "loss": 1.2358,
+      "step": 2555
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.853645915291412e-05,
+      "loss": 1.3074,
+      "step": 2556
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8522869513445515e-05,
+      "loss": 1.3063,
+      "step": 2557
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.850928055966775e-05,
+      "loss": 1.2549,
+      "step": 2558
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8495692297888885e-05,
+      "loss": 1.267,
+      "step": 2559
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.848210473441664e-05,
+      "loss": 1.2571,
+      "step": 2560
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8468517875558433e-05,
+      "loss": 1.2976,
+      "step": 2561
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.845493172762134e-05,
+      "loss": 1.3059,
+      "step": 2562
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8441346296912128e-05,
+      "loss": 1.2603,
+      "step": 2563
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8427761589737203e-05,
+      "loss": 1.2504,
+      "step": 2564
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8414177612402657e-05,
+      "loss": 1.3052,
+      "step": 2565
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8400594371214234e-05,
+      "loss": 1.2742,
+      "step": 2566
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8387011872477338e-05,
+      "loss": 1.2572,
+      "step": 2567
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8373430122497022e-05,
+      "loss": 1.2841,
+      "step": 2568
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.835984912757801e-05,
+      "loss": 1.2668,
+      "step": 2569
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8346268894024644e-05,
+      "loss": 1.2975,
+      "step": 2570
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8332689428140956e-05,
+      "loss": 1.2527,
+      "step": 2571
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.831911073623058e-05,
+      "loss": 1.2708,
+      "step": 2572
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.830553282459681e-05,
+      "loss": 1.2842,
+      "step": 2573
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8291955699542584e-05,
+      "loss": 1.2696,
+      "step": 2574
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8278379367370448e-05,
+      "loss": 1.2579,
+      "step": 2575
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8264803834382622e-05,
+      "loss": 1.284,
+      "step": 2576
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8251229106880916e-05,
+      "loss": 1.2991,
+      "step": 2577
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8237655191166785e-05,
+      "loss": 1.289,
+      "step": 2578
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8224082093541306e-05,
+      "loss": 1.2414,
+      "step": 2579
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8210509820305174e-05,
+      "loss": 1.2379,
+      "step": 2580
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8196938377758696e-05,
+      "loss": 1.2553,
+      "step": 2581
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.818336777220181e-05,
+      "loss": 1.2305,
+      "step": 2582
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8169798009934038e-05,
+      "loss": 1.2905,
+      "step": 2583
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.815622909725454e-05,
+      "loss": 1.3296,
+      "step": 2584
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8142661040462068e-05,
+      "loss": 1.265,
+      "step": 2585
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8129093845854965e-05,
+      "loss": 1.2608,
+      "step": 2586
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.81155275197312e-05,
+      "loss": 1.2616,
+      "step": 2587
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8101962068388315e-05,
+      "loss": 1.2136,
+      "step": 2588
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.808839749812346e-05,
+      "loss": 1.2556,
+      "step": 2589
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.807483381523337e-05,
+      "loss": 1.27,
+      "step": 2590
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8061271026014364e-05,
+      "loss": 1.2541,
+      "step": 2591
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8047709136762368e-05,
+      "loss": 1.2369,
+      "step": 2592
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8034148153772864e-05,
+      "loss": 1.2211,
+      "step": 2593
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8020588083340912e-05,
+      "loss": 1.3083,
+      "step": 2594
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8007028931761184e-05,
+      "loss": 1.2896,
+      "step": 2595
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7993470705327877e-05,
+      "loss": 1.2885,
+      "step": 2596
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.79799134103348e-05,
+      "loss": 1.2292,
+      "step": 2597
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7966357053075312e-05,
+      "loss": 1.299,
+      "step": 2598
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.795280163984232e-05,
+      "loss": 1.2675,
+      "step": 2599
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7939247176928328e-05,
+      "loss": 1.2745,
+      "step": 2600
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.792569367062537e-05,
+      "loss": 1.2909,
+      "step": 2601
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.791214112722505e-05,
+      "loss": 1.2501,
+      "step": 2602
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7898589553018523e-05,
+      "loss": 1.2755,
+      "step": 2603
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.788503895429649e-05,
+      "loss": 1.2249,
+      "step": 2604
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7871489337349208e-05,
+      "loss": 1.2698,
+      "step": 2605
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.785794070846647e-05,
+      "loss": 1.238,
+      "step": 2606
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.78443930739376e-05,
+      "loss": 1.287,
+      "step": 2607
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7830846440051493e-05,
+      "loss": 1.2503,
+      "step": 2608
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7817300813096548e-05,
+      "loss": 1.2666,
+      "step": 2609
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7803756199360704e-05,
+      "loss": 1.2553,
+      "step": 2610
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7790212605131448e-05,
+      "loss": 1.2677,
+      "step": 2611
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7776670036695758e-05,
+      "loss": 1.3103,
+      "step": 2612
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.776312850034018e-05,
+      "loss": 1.2804,
+      "step": 2613
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7749588002350748e-05,
+      "loss": 1.2936,
+      "step": 2614
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7736048549013013e-05,
+      "loss": 1.2828,
+      "step": 2615
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7722510146612075e-05,
+      "loss": 1.1981,
+      "step": 2616
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.770897280143251e-05,
+      "loss": 1.218,
+      "step": 2617
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7695436519758412e-05,
+      "loss": 1.2663,
+      "step": 2618
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.76819013078734e-05,
+      "loss": 1.2229,
+      "step": 2619
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7668367172060562e-05,
+      "loss": 1.284,
+      "step": 2620
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.765483411860253e-05,
+      "loss": 1.308,
+      "step": 2621
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7641302153781402e-05,
+      "loss": 1.2825,
+      "step": 2622
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7627771283878764e-05,
+      "loss": 1.2576,
+      "step": 2623
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.761424151517573e-05,
+      "loss": 1.3091,
+      "step": 2624
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7600712853952863e-05,
+      "loss": 1.2467,
+      "step": 2625
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7587185306490245e-05,
+      "loss": 1.2428,
+      "step": 2626
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7573658879067424e-05,
+      "loss": 1.2758,
+      "step": 2627
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7560133577963423e-05,
+      "loss": 1.2585,
+      "step": 2628
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.754660940945676e-05,
+      "loss": 1.2334,
+      "step": 2629
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.753308637982541e-05,
+      "loss": 1.222,
+      "step": 2630
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.751956449534682e-05,
+      "loss": 1.2623,
+      "step": 2631
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7506043762297932e-05,
+      "loss": 1.32,
+      "step": 2632
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7492524186955108e-05,
+      "loss": 1.2514,
+      "step": 2633
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7479005775594216e-05,
+      "loss": 1.3031,
+      "step": 2634
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.746548853449056e-05,
+      "loss": 1.2556,
+      "step": 2635
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.74519724699189e-05,
+      "loss": 1.2561,
+      "step": 2636
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7438457588153466e-05,
+      "loss": 1.1857,
+      "step": 2637
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.742494389546792e-05,
+      "loss": 1.2086,
+      "step": 2638
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7411431398135384e-05,
+      "loss": 1.2687,
+      "step": 2639
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.739792010242843e-05,
+      "loss": 1.2376,
+      "step": 2640
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.738441001461905e-05,
+      "loss": 1.3122,
+      "step": 2641
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7370901140978706e-05,
+      "loss": 1.2921,
+      "step": 2642
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.735739348777827e-05,
+      "loss": 1.2603,
+      "step": 2643
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.734388706128805e-05,
+      "loss": 1.319,
+      "step": 2644
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7330381867777808e-05,
+      "loss": 1.2287,
+      "step": 2645
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.731687791351671e-05,
+      "loss": 1.2493,
+      "step": 2646
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.730337520477335e-05,
+      "loss": 1.2435,
+      "step": 2647
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7289873747815755e-05,
+      "loss": 1.2438,
+      "step": 2648
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7276373548911355e-05,
+      "loss": 1.2156,
+      "step": 2649
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7262874614327016e-05,
+      "loss": 1.2229,
+      "step": 2650
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7249376950329004e-05,
+      "loss": 1.22,
+      "step": 2651
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7235880563182988e-05,
+      "loss": 1.1856,
+      "step": 2652
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7222385459154072e-05,
+      "loss": 1.3036,
+      "step": 2653
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.720889164450672e-05,
+      "loss": 1.2084,
+      "step": 2654
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7195399125504853e-05,
+      "loss": 1.2874,
+      "step": 2655
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7181907908411744e-05,
+      "loss": 1.2356,
+      "step": 2656
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.716841799949007e-05,
+      "loss": 1.2575,
+      "step": 2657
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7154929405001936e-05,
+      "loss": 1.2283,
+      "step": 2658
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7141442131208788e-05,
+      "loss": 1.2843,
+      "step": 2659
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.712795618437148e-05,
+      "loss": 1.2767,
+      "step": 2660
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7114471570750266e-05,
+      "loss": 1.2471,
+      "step": 2661
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7100988296604756e-05,
+      "loss": 1.2411,
+      "step": 2662
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.708750636819395e-05,
+      "loss": 1.2241,
+      "step": 2663
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7074025791776232e-05,
+      "loss": 1.2435,
+      "step": 2664
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.706054657360933e-05,
+      "loss": 1.2949,
+      "step": 2665
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.704706871995038e-05,
+      "loss": 1.2506,
+      "step": 2666
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.703359223705585e-05,
+      "loss": 1.2397,
+      "step": 2667
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7020117131181585e-05,
+      "loss": 1.3231,
+      "step": 2668
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.70066434085828e-05,
+      "loss": 1.2777,
+      "step": 2669
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6993171075514054e-05,
+      "loss": 1.242,
+      "step": 2670
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.697970013822927e-05,
+      "loss": 1.2717,
+      "step": 2671
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6966230602981727e-05,
+      "loss": 1.2544,
+      "step": 2672
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6952762476024023e-05,
+      "loss": 1.2066,
+      "step": 2673
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6939295763608146e-05,
+      "loss": 1.1992,
+      "step": 2674
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6925830471985398e-05,
+      "loss": 1.2805,
+      "step": 2675
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6912366607406433e-05,
+      "loss": 1.2913,
+      "step": 2676
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6898904176121246e-05,
+      "loss": 1.234,
+      "step": 2677
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.688544318437914e-05,
+      "loss": 1.309,
+      "step": 2678
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6871983638428794e-05,
+      "loss": 1.3139,
+      "step": 2679
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.685852554451818e-05,
+      "loss": 1.2674,
+      "step": 2680
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6845068908894597e-05,
+      "loss": 1.2524,
+      "step": 2681
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.68316137378047e-05,
+      "loss": 1.2966,
+      "step": 2682
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.681816003749442e-05,
+      "loss": 1.2549,
+      "step": 2683
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6804707814209046e-05,
+      "loss": 1.2338,
+      "step": 2684
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6791257074193156e-05,
+      "loss": 1.304,
+      "step": 2685
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.677780782369064e-05,
+      "loss": 1.274,
+      "step": 2686
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6764360068944706e-05,
+      "loss": 1.2277,
+      "step": 2687
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6750913816197873e-05,
+      "loss": 1.3203,
+      "step": 2688
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6737469071691936e-05,
+      "loss": 1.2951,
+      "step": 2689
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6724025841668026e-05,
+      "loss": 1.2688,
+      "step": 2690
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6710584132366542e-05,
+      "loss": 1.288,
+      "step": 2691
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6697143950027194e-05,
+      "loss": 1.2346,
+      "step": 2692
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6683705300888977e-05,
+      "loss": 1.2722,
+      "step": 2693
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.667026819119016e-05,
+      "loss": 1.3016,
+      "step": 2694
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6656832627168338e-05,
+      "loss": 1.3038,
+      "step": 2695
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6643398615060346e-05,
+      "loss": 1.268,
+      "step": 2696
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6629966161102304e-05,
+      "loss": 1.2274,
+      "step": 2697
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.661653527152964e-05,
+      "loss": 1.2902,
+      "step": 2698
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6603105952577024e-05,
+      "loss": 1.2847,
+      "step": 2699
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6589678210478415e-05,
+      "loss": 1.3047,
+      "step": 2700
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.657625205146703e-05,
+      "loss": 1.2448,
+      "step": 2701
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6562827481775353e-05,
+      "loss": 1.285,
+      "step": 2702
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6549404507635135e-05,
+      "loss": 1.2517,
+      "step": 2703
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6535983135277378e-05,
+      "loss": 1.316,
+      "step": 2704
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6522563370932355e-05,
+      "loss": 1.2458,
+      "step": 2705
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6509145220829574e-05,
+      "loss": 1.2566,
+      "step": 2706
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.64957286911978e-05,
+      "loss": 1.2483,
+      "step": 2707
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6482313788265058e-05,
+      "loss": 1.2282,
+      "step": 2708
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.64689005182586e-05,
+      "loss": 1.2725,
+      "step": 2709
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6455488887404935e-05,
+      "loss": 1.2895,
+      "step": 2710
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6442078901929803e-05,
+      "loss": 1.2979,
+      "step": 2711
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6428670568058176e-05,
+      "loss": 1.2606,
+      "step": 2712
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.641526389201427e-05,
+      "loss": 1.2355,
+      "step": 2713
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.640185888002153e-05,
+      "loss": 1.285,
+      "step": 2714
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6388455538302612e-05,
+      "loss": 1.2582,
+      "step": 2715
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6375053873079424e-05,
+      "loss": 1.2936,
+      "step": 2716
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6361653890573078e-05,
+      "loss": 1.2875,
+      "step": 2717
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6348255597003896e-05,
+      "loss": 1.2882,
+      "step": 2718
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.633485899859144e-05,
+      "loss": 1.2174,
+      "step": 2719
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.632146410155447e-05,
+      "loss": 1.1721,
+      "step": 2720
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6308070912110965e-05,
+      "loss": 1.2582,
+      "step": 2721
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6294679436478095e-05,
+      "loss": 1.2698,
+      "step": 2722
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6281289680872252e-05,
+      "loss": 1.2157,
+      "step": 2723
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6267901651509022e-05,
+      "loss": 1.2678,
+      "step": 2724
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6254515354603194e-05,
+      "loss": 1.2412,
+      "step": 2725
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6241130796368737e-05,
+      "loss": 1.2696,
+      "step": 2726
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6227747983018845e-05,
+      "loss": 1.2756,
+      "step": 2727
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6214366920765856e-05,
+      "loss": 1.2865,
+      "step": 2728
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.620098761582135e-05,
+      "loss": 1.2554,
+      "step": 2729
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6187610074396044e-05,
+      "loss": 1.2372,
+      "step": 2730
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6174234302699856e-05,
+      "loss": 1.291,
+      "step": 2731
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.616086030694189e-05,
+      "loss": 1.2583,
+      "step": 2732
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6147488093330405e-05,
+      "loss": 1.2837,
+      "step": 2733
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6134117668072858e-05,
+      "loss": 1.2748,
+      "step": 2734
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.612074903737585e-05,
+      "loss": 1.2685,
+      "step": 2735
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.610738220744517e-05,
+      "loss": 1.2561,
+      "step": 2736
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6094017184485763e-05,
+      "loss": 1.2569,
+      "step": 2737
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6080653974701732e-05,
+      "loss": 1.2776,
+      "step": 2738
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6067292584296333e-05,
+      "loss": 1.2768,
+      "step": 2739
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6053933019472003e-05,
+      "loss": 1.2387,
+      "step": 2740
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6040575286430295e-05,
+      "loss": 1.2502,
+      "step": 2741
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.602721939137195e-05,
+      "loss": 1.2123,
+      "step": 2742
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6013865340496826e-05,
+      "loss": 1.2785,
+      "step": 2743
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6000513140003927e-05,
+      "loss": 1.2757,
+      "step": 2744
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5987162796091428e-05,
+      "loss": 1.2975,
+      "step": 2745
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5973814314956602e-05,
+      "loss": 1.2783,
+      "step": 2746
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.596046770279588e-05,
+      "loss": 1.2931,
+      "step": 2747
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5947122965804827e-05,
+      "loss": 1.3106,
+      "step": 2748
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5933780110178128e-05,
+      "loss": 1.2796,
+      "step": 2749
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.59204391421096e-05,
+      "loss": 1.2381,
+      "step": 2750
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5907100067792186e-05,
+      "loss": 1.2646,
+      "step": 2751
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.589376289341793e-05,
+      "loss": 1.266,
+      "step": 2752
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5880427625178035e-05,
+      "loss": 1.2513,
+      "step": 2753
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.586709426926277e-05,
+      "loss": 1.2819,
+      "step": 2754
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5853762831861567e-05,
+      "loss": 1.2558,
+      "step": 2755
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5840433319162925e-05,
+      "loss": 1.2937,
+      "step": 2756
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5827105737354456e-05,
+      "loss": 1.2824,
+      "step": 2757
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5813780092622907e-05,
+      "loss": 1.261,
+      "step": 2758
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.580045639115409e-05,
+      "loss": 1.2662,
+      "step": 2759
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5787134639132935e-05,
+      "loss": 1.3024,
+      "step": 2760
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.577381484274346e-05,
+      "loss": 1.2585,
+      "step": 2761
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.576049700816877e-05,
+      "loss": 1.2769,
+      "step": 2762
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.574718114159108e-05,
+      "loss": 1.2846,
+      "step": 2763
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5733867249191667e-05,
+      "loss": 1.2085,
+      "step": 2764
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.57205553371509e-05,
+      "loss": 1.2509,
+      "step": 2765
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.570724541164824e-05,
+      "loss": 1.2795,
+      "step": 2766
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.569393747886221e-05,
+      "loss": 1.2655,
+      "step": 2767
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5680631544970405e-05,
+      "loss": 1.2706,
+      "step": 2768
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5667327616149522e-05,
+      "loss": 1.2338,
+      "step": 2769
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5654025698575286e-05,
+      "loss": 1.3053,
+      "step": 2770
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5640725798422525e-05,
+      "loss": 1.2229,
+      "step": 2771
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5627427921865106e-05,
+      "loss": 1.2047,
+      "step": 2772
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5614132075075967e-05,
+      "loss": 1.2788,
+      "step": 2773
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5600838264227102e-05,
+      "loss": 1.2995,
+      "step": 2774
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5587546495489563e-05,
+      "loss": 1.2613,
+      "step": 2775
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.557425677503344e-05,
+      "loss": 1.2044,
+      "step": 2776
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5560969109027896e-05,
+      "loss": 1.2386,
+      "step": 2777
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5547683503641115e-05,
+      "loss": 1.2889,
+      "step": 2778
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5534399965040353e-05,
+      "loss": 1.2513,
+      "step": 2779
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.552111849939188e-05,
+      "loss": 1.2381,
+      "step": 2780
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.550783911286101e-05,
+      "loss": 1.2871,
+      "step": 2781
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5494561811612102e-05,
+      "loss": 1.3133,
+      "step": 2782
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.548128660180854e-05,
+      "loss": 1.2307,
+      "step": 2783
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5468013489612742e-05,
+      "loss": 1.2312,
+      "step": 2784
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5454742481186137e-05,
+      "loss": 1.3291,
+      "step": 2785
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5441473582689198e-05,
+      "loss": 1.2897,
+      "step": 2786
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5428206800281413e-05,
+      "loss": 1.2645,
+      "step": 2787
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5414942140121278e-05,
+      "loss": 1.2757,
+      "step": 2788
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.54016796083663e-05,
+      "loss": 1.2955,
+      "step": 2789
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.538841921117303e-05,
+      "loss": 1.2767,
+      "step": 2790
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5375160954696986e-05,
+      "loss": 1.219,
+      "step": 2791
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.536190484509273e-05,
+      "loss": 1.2828,
+      "step": 2792
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5348650888513798e-05,
+      "loss": 1.1934,
+      "step": 2793
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.533539909111273e-05,
+      "loss": 1.2265,
+      "step": 2794
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5322149459041097e-05,
+      "loss": 1.2764,
+      "step": 2795
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5308901998449415e-05,
+      "loss": 1.232,
+      "step": 2796
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5295656715487226e-05,
+      "loss": 1.2899,
+      "step": 2797
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5282413616303063e-05,
+      "loss": 1.3017,
+      "step": 2798
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.526917270704441e-05,
+      "loss": 1.2194,
+      "step": 2799
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5255933993857785e-05,
+      "loss": 1.2998,
+      "step": 2800
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5242697482888649e-05,
+      "loss": 1.2646,
+      "step": 2801
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5229463180281441e-05,
+      "loss": 1.2445,
+      "step": 2802
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5216231092179604e-05,
+      "loss": 1.2482,
+      "step": 2803
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5203001224725525e-05,
+      "loss": 1.2311,
+      "step": 2804
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5189773584060563e-05,
+      "loss": 1.2904,
+      "step": 2805
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.517654817632507e-05,
+      "loss": 1.2851,
+      "step": 2806
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5163325007658319e-05,
+      "loss": 1.2917,
+      "step": 2807
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5150104084198587e-05,
+      "loss": 1.2033,
+      "step": 2808
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5136885412083073e-05,
+      "loss": 1.255,
+      "step": 2809
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5123668997447948e-05,
+      "loss": 1.2871,
+      "step": 2810
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5110454846428348e-05,
+      "loss": 1.2719,
+      "step": 2811
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5097242965158322e-05,
+      "loss": 1.246,
+      "step": 2812
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5084033359770907e-05,
+      "loss": 1.2591,
+      "step": 2813
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5070826036398052e-05,
+      "loss": 1.2777,
+      "step": 2814
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.505762100117065e-05,
+      "loss": 1.1695,
+      "step": 2815
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5044418260218559e-05,
+      "loss": 1.3016,
+      "step": 2816
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.503121781967054e-05,
+      "loss": 1.3046,
+      "step": 2817
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5018019685654295e-05,
+      "loss": 1.2725,
+      "step": 2818
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5004823864296472e-05,
+      "loss": 1.2431,
+      "step": 2819
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4991630361722619e-05,
+      "loss": 1.2256,
+      "step": 2820
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4978439184057233e-05,
+      "loss": 1.202,
+      "step": 2821
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4965250337423718e-05,
+      "loss": 1.2295,
+      "step": 2822
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4952063827944385e-05,
+      "loss": 1.283,
+      "step": 2823
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4938879661740495e-05,
+      "loss": 1.2384,
+      "step": 2824
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4925697844932185e-05,
+      "loss": 1.2484,
+      "step": 2825
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4912518383638512e-05,
+      "loss": 1.2424,
+      "step": 2826
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4899341283977457e-05,
+      "loss": 1.2382,
+      "step": 2827
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4886166552065873e-05,
+      "loss": 1.2815,
+      "step": 2828
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4872994194019553e-05,
+      "loss": 1.2739,
+      "step": 2829
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4859824215953154e-05,
+      "loss": 1.2352,
+      "step": 2830
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4846656623980234e-05,
+      "loss": 1.211,
+      "step": 2831
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4833491424213268e-05,
+      "loss": 1.1948,
+      "step": 2832
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4820328622763584e-05,
+      "loss": 1.2689,
+      "step": 2833
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4807168225741433e-05,
+      "loss": 1.2698,
+      "step": 2834
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4794010239255925e-05,
+      "loss": 1.2735,
+      "step": 2835
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4780854669415053e-05,
+      "loss": 1.2473,
+      "step": 2836
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4767701522325708e-05,
+      "loss": 1.3154,
+      "step": 2837
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4754550804093633e-05,
+      "loss": 1.228,
+      "step": 2838
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4741402520823442e-05,
+      "loss": 1.2206,
+      "step": 2839
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4728256678618652e-05,
+      "loss": 1.2807,
+      "step": 2840
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.47151132835816e-05,
+      "loss": 1.2791,
+      "step": 2841
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4701972341813533e-05,
+      "loss": 1.23,
+      "step": 2842
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4688833859414529e-05,
+      "loss": 1.2019,
+      "step": 2843
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.467569784248352e-05,
+      "loss": 1.263,
+      "step": 2844
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4662564297118325e-05,
+      "loss": 1.2614,
+      "step": 2845
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4649433229415588e-05,
+      "loss": 1.2007,
+      "step": 2846
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4636304645470807e-05,
+      "loss": 1.2403,
+      "step": 2847
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4623178551378346e-05,
+      "loss": 1.2279,
+      "step": 2848
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4610054953231379e-05,
+      "loss": 1.2979,
+      "step": 2849
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4596933857121963e-05,
+      "loss": 1.2706,
+      "step": 2850
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4583815269140957e-05,
+      "loss": 1.2874,
+      "step": 2851
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4570699195378071e-05,
+      "loss": 1.2471,
+      "step": 2852
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4557585641921859e-05,
+      "loss": 1.2704,
+      "step": 2853
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4544474614859683e-05,
+      "loss": 1.2611,
+      "step": 2854
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4531366120277736e-05,
+      "loss": 1.2439,
+      "step": 2855
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4518260164261058e-05,
+      "loss": 1.273,
+      "step": 2856
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4505156752893488e-05,
+      "loss": 1.2756,
+      "step": 2857
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4492055892257688e-05,
+      "loss": 1.2911,
+      "step": 2858
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4478957588435148e-05,
+      "loss": 1.2161,
+      "step": 2859
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4465861847506142e-05,
+      "loss": 1.2703,
+      "step": 2860
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4452768675549798e-05,
+      "loss": 1.2377,
+      "step": 2861
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4439678078644004e-05,
+      "loss": 1.2483,
+      "step": 2862
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4426590062865497e-05,
+      "loss": 1.242,
+      "step": 2863
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4413504634289785e-05,
+      "loss": 1.2969,
+      "step": 2864
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4400421798991178e-05,
+      "loss": 1.2291,
+      "step": 2865
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4387341563042801e-05,
+      "loss": 1.2642,
+      "step": 2866
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4374263932516557e-05,
+      "loss": 1.3157,
+      "step": 2867
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4361188913483132e-05,
+      "loss": 1.2582,
+      "step": 2868
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4348116512012024e-05,
+      "loss": 1.2395,
+      "step": 2869
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4335046734171499e-05,
+      "loss": 1.2779,
+      "step": 2870
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4321979586028607e-05,
+      "loss": 1.2751,
+      "step": 2871
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4308915073649182e-05,
+      "loss": 1.2613,
+      "step": 2872
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4295853203097823e-05,
+      "loss": 1.2756,
+      "step": 2873
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4282793980437923e-05,
+      "loss": 1.2659,
+      "step": 2874
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4269737411731627e-05,
+      "loss": 1.3003,
+      "step": 2875
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.425668350303985e-05,
+      "loss": 1.237,
+      "step": 2876
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4243632260422292e-05,
+      "loss": 1.2457,
+      "step": 2877
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4230583689937381e-05,
+      "loss": 1.2737,
+      "step": 2878
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4217537797642343e-05,
+      "loss": 1.2034,
+      "step": 2879
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4204494589593127e-05,
+      "loss": 1.2646,
+      "step": 2880
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4191454071844457e-05,
+      "loss": 1.2501,
+      "step": 2881
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4178416250449804e-05,
+      "loss": 1.2616,
+      "step": 2882
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4165381131461388e-05,
+      "loss": 1.2314,
+      "step": 2883
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4152348720930156e-05,
+      "loss": 1.2869,
+      "step": 2884
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4139319024905836e-05,
+      "loss": 1.2607,
+      "step": 2885
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.412629204943685e-05,
+      "loss": 1.2817,
+      "step": 2886
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4113267800570402e-05,
+      "loss": 1.2456,
+      "step": 2887
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.41002462843524e-05,
+      "loss": 1.2291,
+      "step": 2888
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4087227506827482e-05,
+      "loss": 1.2301,
+      "step": 2889
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4074211474039046e-05,
+      "loss": 1.2349,
+      "step": 2890
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.406119819202917e-05,
+      "loss": 1.2626,
+      "step": 2891
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4048187666838707e-05,
+      "loss": 1.2817,
+      "step": 2892
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4035179904507184e-05,
+      "loss": 1.3066,
+      "step": 2893
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4022174911072868e-05,
+      "loss": 1.2531,
+      "step": 2894
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4009172692572743e-05,
+      "loss": 1.2722,
+      "step": 2895
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.39961732550425e-05,
+      "loss": 1.2629,
+      "step": 2896
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3983176604516526e-05,
+      "loss": 1.2283,
+      "step": 2897
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3970182747027944e-05,
+      "loss": 1.3009,
+      "step": 2898
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3957191688608544e-05,
+      "loss": 1.2429,
+      "step": 2899
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3944203435288857e-05,
+      "loss": 1.2579,
+      "step": 2900
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3931217993098076e-05,
+      "loss": 1.2907,
+      "step": 2901
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3918235368064102e-05,
+      "loss": 1.2069,
+      "step": 2902
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3905255566213542e-05,
+      "loss": 1.2855,
+      "step": 2903
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3892278593571669e-05,
+      "loss": 1.2529,
+      "step": 2904
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3879304456162457e-05,
+      "loss": 1.215,
+      "step": 2905
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3866333160008562e-05,
+      "loss": 1.2576,
+      "step": 2906
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3853364711131324e-05,
+      "loss": 1.1791,
+      "step": 2907
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3840399115550748e-05,
+      "loss": 1.2156,
+      "step": 2908
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3827436379285537e-05,
+      "loss": 1.2298,
+      "step": 2909
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3814476508353036e-05,
+      "loss": 1.2701,
+      "step": 2910
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3801519508769295e-05,
+      "loss": 1.253,
+      "step": 2911
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3788565386548996e-05,
+      "loss": 1.2261,
+      "step": 2912
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3775614147705521e-05,
+      "loss": 1.2228,
+      "step": 2913
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3762665798250887e-05,
+      "loss": 1.263,
+      "step": 2914
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3749720344195768e-05,
+      "loss": 1.243,
+      "step": 2915
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.373677779154952e-05,
+      "loss": 1.2675,
+      "step": 2916
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3723838146320128e-05,
+      "loss": 1.2731,
+      "step": 2917
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3710901414514235e-05,
+      "loss": 1.2577,
+      "step": 2918
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3697967602137135e-05,
+      "loss": 1.3023,
+      "step": 2919
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.368503671519276e-05,
+      "loss": 1.2115,
+      "step": 2920
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3672108759683694e-05,
+      "loss": 1.2979,
+      "step": 2921
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3659183741611154e-05,
+      "loss": 1.2026,
+      "step": 2922
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3646261666974976e-05,
+      "loss": 1.2819,
+      "step": 2923
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3633342541773673e-05,
+      "loss": 1.2395,
+      "step": 2924
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3620426372004353e-05,
+      "loss": 1.2855,
+      "step": 2925
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.360751316366275e-05,
+      "loss": 1.2285,
+      "step": 2926
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3594602922743252e-05,
+      "loss": 1.2752,
+      "step": 2927
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.358169565523884e-05,
+      "loss": 1.2844,
+      "step": 2928
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.356879136714114e-05,
+      "loss": 1.2648,
+      "step": 2929
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3555890064440374e-05,
+      "loss": 1.2449,
+      "step": 2930
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3542991753125387e-05,
+      "loss": 1.2562,
+      "step": 2931
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3530096439183637e-05,
+      "loss": 1.2648,
+      "step": 2932
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3517204128601193e-05,
+      "loss": 1.3028,
+      "step": 2933
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3504314827362715e-05,
+      "loss": 1.2734,
+      "step": 2934
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3491428541451487e-05,
+      "loss": 1.2486,
+      "step": 2935
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3478545276849373e-05,
+      "loss": 1.2236,
+      "step": 2936
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3465665039536857e-05,
+      "loss": 1.2125,
+      "step": 2937
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3452787835492998e-05,
+      "loss": 1.2841,
+      "step": 2938
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3439913670695445e-05,
+      "loss": 1.272,
+      "step": 2939
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3427042551120461e-05,
+      "loss": 1.2461,
+      "step": 2940
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3414174482742865e-05,
+      "loss": 1.2438,
+      "step": 2941
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3401309471536092e-05,
+      "loss": 1.3218,
+      "step": 2942
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3388447523472122e-05,
+      "loss": 1.2427,
+      "step": 2943
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.337558864452154e-05,
+      "loss": 1.2186,
+      "step": 2944
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3362732840653494e-05,
+      "loss": 1.281,
+      "step": 2945
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3349880117835716e-05,
+      "loss": 1.2928,
+      "step": 2946
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3337030482034485e-05,
+      "loss": 1.238,
+      "step": 2947
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.332418393921468e-05,
+      "loss": 1.2709,
+      "step": 2948
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3311340495339704e-05,
+      "loss": 1.2721,
+      "step": 2949
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3298500156371565e-05,
+      "loss": 1.2556,
+      "step": 2950
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.32856629282708e-05,
+      "loss": 1.2359,
+      "step": 2951
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3272828816996498e-05,
+      "loss": 1.1736,
+      "step": 2952
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3259997828506333e-05,
+      "loss": 1.236,
+      "step": 2953
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3247169968756494e-05,
+      "loss": 1.2504,
+      "step": 2954
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3234345243701743e-05,
+      "loss": 1.2845,
+      "step": 2955
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.322152365929537e-05,
+      "loss": 1.3128,
+      "step": 2956
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.320870522148921e-05,
+      "loss": 1.2583,
+      "step": 2957
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3195889936233662e-05,
+      "loss": 1.2285,
+      "step": 2958
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3183077809477625e-05,
+      "loss": 1.244,
+      "step": 2959
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3170268847168541e-05,
+      "loss": 1.2554,
+      "step": 2960
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.315746305525241e-05,
+      "loss": 1.2997,
+      "step": 2961
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3144660439673727e-05,
+      "loss": 1.2706,
+      "step": 2962
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3131861006375519e-05,
+      "loss": 1.2711,
+      "step": 2963
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.311906476129936e-05,
+      "loss": 1.2799,
+      "step": 2964
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3106271710385312e-05,
+      "loss": 1.2707,
+      "step": 2965
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3093481859571981e-05,
+      "loss": 1.2544,
+      "step": 2966
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3080695214796464e-05,
+      "loss": 1.3039,
+      "step": 2967
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3067911781994384e-05,
+      "loss": 1.268,
+      "step": 2968
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3055131567099872e-05,
+      "loss": 1.2769,
+      "step": 2969
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3042354576045559e-05,
+      "loss": 1.2833,
+      "step": 2970
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3029580814762596e-05,
+      "loss": 1.3312,
+      "step": 2971
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.3016810289180615e-05,
+      "loss": 1.2674,
+      "step": 2972
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.3004043005227742e-05,
+      "loss": 1.2727,
+      "step": 2973
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2991278968830625e-05,
+      "loss": 1.2411,
+      "step": 2974
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2978518185914388e-05,
+      "loss": 1.2056,
+      "step": 2975
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.296576066240263e-05,
+      "loss": 1.2812,
+      "step": 2976
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2953006404217474e-05,
+      "loss": 1.2758,
+      "step": 2977
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2940255417279486e-05,
+      "loss": 1.2314,
+      "step": 2978
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2927507707507751e-05,
+      "loss": 1.2708,
+      "step": 2979
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2914763280819804e-05,
+      "loss": 1.2673,
+      "step": 2980
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2902022143131668e-05,
+      "loss": 1.2948,
+      "step": 2981
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2889284300357847e-05,
+      "loss": 1.19,
+      "step": 2982
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.28765497584113e-05,
+      "loss": 1.2836,
+      "step": 2983
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2863818523203452e-05,
+      "loss": 1.2801,
+      "step": 2984
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2851090600644223e-05,
+      "loss": 1.2242,
+      "step": 2985
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2838365996641949e-05,
+      "loss": 1.214,
+      "step": 2986
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2825644717103472e-05,
+      "loss": 1.2517,
+      "step": 2987
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.281292676793406e-05,
+      "loss": 1.2531,
+      "step": 2988
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2800212155037437e-05,
+      "loss": 1.2424,
+      "step": 2989
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.27875008843158e-05,
+      "loss": 1.2163,
+      "step": 2990
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2774792961669764e-05,
+      "loss": 1.2368,
+      "step": 2991
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2762088392998417e-05,
+      "loss": 1.2562,
+      "step": 2992
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2749387184199283e-05,
+      "loss": 1.2586,
+      "step": 2993
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2736689341168304e-05,
+      "loss": 1.2376,
+      "step": 2994
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2723994869799898e-05,
+      "loss": 1.276,
+      "step": 2995
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2711303775986888e-05,
+      "loss": 1.2508,
+      "step": 2996
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2698616065620528e-05,
+      "loss": 1.2692,
+      "step": 2997
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2685931744590536e-05,
+      "loss": 1.2778,
+      "step": 2998
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.267325081878501e-05,
+      "loss": 1.2686,
+      "step": 2999
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2660573294090512e-05,
+      "loss": 1.1985,
+      "step": 3000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2647899176392003e-05,
+      "loss": 1.2766,
+      "step": 3001
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2635228471572853e-05,
+      "loss": 1.2287,
+      "step": 3002
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2622561185514886e-05,
+      "loss": 1.2832,
+      "step": 3003
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2609897324098297e-05,
+      "loss": 1.2815,
+      "step": 3004
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2597236893201712e-05,
+      "loss": 1.2282,
+      "step": 3005
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2584579898702175e-05,
+      "loss": 1.2733,
+      "step": 3006
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.25719263464751e-05,
+      "loss": 1.2478,
+      "step": 3007
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2559276242394347e-05,
+      "loss": 1.2505,
+      "step": 3008
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.254662959233214e-05,
+      "loss": 1.2536,
+      "step": 3009
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2533986402159113e-05,
+      "loss": 1.226,
+      "step": 3010
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.25213466777443e-05,
+      "loss": 1.2879,
+      "step": 3011
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2508710424955117e-05,
+      "loss": 1.2379,
+      "step": 3012
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.249607764965736e-05,
+      "loss": 1.2265,
+      "step": 3013
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2483448357715242e-05,
+      "loss": 1.2394,
+      "step": 3014
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2470822554991321e-05,
+      "loss": 1.2516,
+      "step": 3015
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2458200247346569e-05,
+      "loss": 1.2821,
+      "step": 3016
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2445581440640312e-05,
+      "loss": 1.2852,
+      "step": 3017
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.243296614073025e-05,
+      "loss": 1.1998,
+      "step": 3018
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2420354353472483e-05,
+      "loss": 1.25,
+      "step": 3019
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2407746084721444e-05,
+      "loss": 1.2859,
+      "step": 3020
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2395141340329966e-05,
+      "loss": 1.2565,
+      "step": 3021
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2382540126149218e-05,
+      "loss": 1.2491,
+      "step": 3022
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2369942448028738e-05,
+      "loss": 1.2762,
+      "step": 3023
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2357348311816444e-05,
+      "loss": 1.2898,
+      "step": 3024
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2344757723358583e-05,
+      "loss": 1.2155,
+      "step": 3025
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2332170688499753e-05,
+      "loss": 1.2094,
+      "step": 3026
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2319587213082931e-05,
+      "loss": 1.2514,
+      "step": 3027
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.230700730294942e-05,
+      "loss": 1.2304,
+      "step": 3028
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.229443096393887e-05,
+      "loss": 1.2681,
+      "step": 3029
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2281858201889283e-05,
+      "loss": 1.2783,
+      "step": 3030
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.226928902263698e-05,
+      "loss": 1.2335,
+      "step": 3031
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2256723432016648e-05,
+      "loss": 1.2521,
+      "step": 3032
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2244161435861286e-05,
+      "loss": 1.2283,
+      "step": 3033
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2231603040002225e-05,
+      "loss": 1.2738,
+      "step": 3034
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2219048250269141e-05,
+      "loss": 1.2477,
+      "step": 3035
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2206497072490014e-05,
+      "loss": 1.2762,
+      "step": 3036
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2193949512491172e-05,
+      "loss": 1.1641,
+      "step": 3037
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2181405576097247e-05,
+      "loss": 1.2404,
+      "step": 3038
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2168865269131182e-05,
+      "loss": 1.2503,
+      "step": 3039
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2156328597414258e-05,
+      "loss": 1.2949,
+      "step": 3040
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2143795566766054e-05,
+      "loss": 1.2353,
+      "step": 3041
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2131266183004455e-05,
+      "loss": 1.2002,
+      "step": 3042
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2118740451945668e-05,
+      "loss": 1.3042,
+      "step": 3043
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2106218379404187e-05,
+      "loss": 1.2446,
+      "step": 3044
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2093699971192825e-05,
+      "loss": 1.2387,
+      "step": 3045
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.208118523312268e-05,
+      "loss": 1.2728,
+      "step": 3046
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2068674171003146e-05,
+      "loss": 1.2408,
+      "step": 3047
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.205616679064193e-05,
+      "loss": 1.2971,
+      "step": 3048
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2043663097844999e-05,
+      "loss": 1.225,
+      "step": 3049
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2031163098416644e-05,
+      "loss": 1.2419,
+      "step": 3050
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2018666798159408e-05,
+      "loss": 1.2404,
+      "step": 3051
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2006174202874141e-05,
+      "loss": 1.2231,
+      "step": 3052
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1993685318359956e-05,
+      "loss": 1.2834,
+      "step": 3053
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1981200150414262e-05,
+      "loss": 1.2667,
+      "step": 3054
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1968718704832716e-05,
+      "loss": 1.2504,
+      "step": 3055
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.195624098740928e-05,
+      "loss": 1.3046,
+      "step": 3056
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1943767003936152e-05,
+      "loss": 1.3131,
+      "step": 3057
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1931296760203831e-05,
+      "loss": 1.1847,
+      "step": 3058
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1918830262001055e-05,
+      "loss": 1.2474,
+      "step": 3059
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1906367515114816e-05,
+      "loss": 1.2154,
+      "step": 3060
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1893908525330401e-05,
+      "loss": 1.2043,
+      "step": 3061
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1881453298431323e-05,
+      "loss": 1.2553,
+      "step": 3062
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1869001840199345e-05,
+      "loss": 1.2315,
+      "step": 3063
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1856554156414503e-05,
+      "loss": 1.2944,
+      "step": 3064
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1844110252855072e-05,
+      "loss": 1.2535,
+      "step": 3065
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1831670135297564e-05,
+      "loss": 1.2273,
+      "step": 3066
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1819233809516746e-05,
+      "loss": 1.2651,
+      "step": 3067
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1806801281285604e-05,
+      "loss": 1.2971,
+      "step": 3068
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1794372556375392e-05,
+      "loss": 1.1915,
+      "step": 3069
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.178194764055557e-05,
+      "loss": 1.2387,
+      "step": 3070
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.176952653959385e-05,
+      "loss": 1.2562,
+      "step": 3071
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1757109259256163e-05,
+      "loss": 1.2709,
+      "step": 3072
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.174469580530666e-05,
+      "loss": 1.2709,
+      "step": 3073
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1732286183507738e-05,
+      "loss": 1.2748,
+      "step": 3074
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1719880399619987e-05,
+      "loss": 1.2853,
+      "step": 3075
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1707478459402236e-05,
+      "loss": 1.2439,
+      "step": 3076
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1695080368611526e-05,
+      "loss": 1.2483,
+      "step": 3077
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1682686133003105e-05,
+      "loss": 1.2746,
+      "step": 3078
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.167029575833044e-05,
+      "loss": 1.2671,
+      "step": 3079
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1657909250345194e-05,
+      "loss": 1.262,
+      "step": 3080
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1645526614797235e-05,
+      "loss": 1.233,
+      "step": 3081
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1633147857434658e-05,
+      "loss": 1.2542,
+      "step": 3082
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1620772984003724e-05,
+      "loss": 1.1922,
+      "step": 3083
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1608402000248908e-05,
+      "loss": 1.2216,
+      "step": 3084
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1596034911912896e-05,
+      "loss": 1.2636,
+      "step": 3085
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1583671724736526e-05,
+      "loss": 1.2652,
+      "step": 3086
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.157131244445886e-05,
+      "loss": 1.2796,
+      "step": 3087
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1558957076817135e-05,
+      "loss": 1.2416,
+      "step": 3088
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1546605627546752e-05,
+      "loss": 1.2384,
+      "step": 3089
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1534258102381332e-05,
+      "loss": 1.2593,
+      "step": 3090
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1521914507052646e-05,
+      "loss": 1.2582,
+      "step": 3091
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1509574847290641e-05,
+      "loss": 1.2765,
+      "step": 3092
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1497239128823456e-05,
+      "loss": 1.2534,
+      "step": 3093
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1484907357377378e-05,
+      "loss": 1.2489,
+      "step": 3094
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1472579538676883e-05,
+      "loss": 1.2354,
+      "step": 3095
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1460255678444598e-05,
+      "loss": 1.2463,
+      "step": 3096
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.144793578240131e-05,
+      "loss": 1.2898,
+      "step": 3097
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1435619856265982e-05,
+      "loss": 1.2528,
+      "step": 3098
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1423307905755714e-05,
+      "loss": 1.2369,
+      "step": 3099
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1410999936585782e-05,
+      "loss": 1.244,
+      "step": 3100
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1398695954469598e-05,
+      "loss": 1.2379,
+      "step": 3101
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1386395965118715e-05,
+      "loss": 1.2566,
+      "step": 3102
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1374099974242867e-05,
+      "loss": 1.208,
+      "step": 3103
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.13618079875499e-05,
+      "loss": 1.2772,
+      "step": 3104
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1349520010745802e-05,
+      "loss": 1.2572,
+      "step": 3105
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1337236049534726e-05,
+      "loss": 1.2293,
+      "step": 3106
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1324956109618927e-05,
+      "loss": 1.2289,
+      "step": 3107
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1312680196698817e-05,
+      "loss": 1.2304,
+      "step": 3108
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1300408316472944e-05,
+      "loss": 1.2079,
+      "step": 3109
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1288140474637953e-05,
+      "loss": 1.2966,
+      "step": 3110
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.127587667688865e-05,
+      "loss": 1.2413,
+      "step": 3111
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.126361692891794e-05,
+      "loss": 1.2613,
+      "step": 3112
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1251361236416845e-05,
+      "loss": 1.2146,
+      "step": 3113
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1239109605074527e-05,
+      "loss": 1.2915,
+      "step": 3114
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1226862040578244e-05,
+      "loss": 1.2645,
+      "step": 3115
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1214618548613379e-05,
+      "loss": 1.2736,
+      "step": 3116
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1202379134863412e-05,
+      "loss": 1.2165,
+      "step": 3117
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1190143805009934e-05,
+      "loss": 1.2914,
+      "step": 3118
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.117791256473265e-05,
+      "loss": 1.2711,
+      "step": 3119
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1165685419709353e-05,
+      "loss": 1.2913,
+      "step": 3120
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1153462375615934e-05,
+      "loss": 1.2376,
+      "step": 3121
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1141243438126403e-05,
+      "loss": 1.2159,
+      "step": 3122
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1129028612912832e-05,
+      "loss": 1.2791,
+      "step": 3123
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1116817905645411e-05,
+      "loss": 1.2404,
+      "step": 3124
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1104611321992404e-05,
+      "loss": 1.1767,
+      "step": 3125
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1092408867620155e-05,
+      "loss": 1.2579,
+      "step": 3126
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1080210548193113e-05,
+      "loss": 1.2654,
+      "step": 3127
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1068016369373784e-05,
+      "loss": 1.2605,
+      "step": 3128
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1055826336822775e-05,
+      "loss": 1.2506,
+      "step": 3129
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1043640456198745e-05,
+      "loss": 1.2704,
+      "step": 3130
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1031458733158434e-05,
+      "loss": 1.2567,
+      "step": 3131
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.101928117335666e-05,
+      "loss": 1.2427,
+      "step": 3132
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.100710778244631e-05,
+      "loss": 1.3009,
+      "step": 3133
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0994938566078315e-05,
+      "loss": 1.333,
+      "step": 3134
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0982773529901696e-05,
+      "loss": 1.2514,
+      "step": 3135
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0970612679563501e-05,
+      "loss": 1.2851,
+      "step": 3136
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0958456020708875e-05,
+      "loss": 1.2252,
+      "step": 3137
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0946303558980981e-05,
+      "loss": 1.2286,
+      "step": 3138
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0934155300021048e-05,
+      "loss": 1.2222,
+      "step": 3139
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0922011249468362e-05,
+      "loss": 1.276,
+      "step": 3140
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0909871412960245e-05,
+      "loss": 1.2708,
+      "step": 3141
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0897735796132056e-05,
+      "loss": 1.2226,
+      "step": 3142
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0885604404617221e-05,
+      "loss": 1.2381,
+      "step": 3143
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.087347724404717e-05,
+      "loss": 1.2114,
+      "step": 3144
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.086135432005141e-05,
+      "loss": 1.2417,
+      "step": 3145
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0849235638257442e-05,
+      "loss": 1.2634,
+      "step": 3146
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0837121204290812e-05,
+      "loss": 1.2382,
+      "step": 3147
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0825011023775111e-05,
+      "loss": 1.2725,
+      "step": 3148
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0812905102331927e-05,
+      "loss": 1.2429,
+      "step": 3149
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0800803445580896e-05,
+      "loss": 1.2722,
+      "step": 3150
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.078870605913966e-05,
+      "loss": 1.2428,
+      "step": 3151
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0776612948623874e-05,
+      "loss": 1.2101,
+      "step": 3152
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0764524119647228e-05,
+      "loss": 1.2547,
+      "step": 3153
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0752439577821398e-05,
+      "loss": 1.2091,
+      "step": 3154
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0740359328756105e-05,
+      "loss": 1.2598,
+      "step": 3155
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0728283378059036e-05,
+      "loss": 1.2172,
+      "step": 3156
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0716211731335922e-05,
+      "loss": 1.2211,
+      "step": 3157
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0704144394190458e-05,
+      "loss": 1.2192,
+      "step": 3158
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0692081372224378e-05,
+      "loss": 1.2634,
+      "step": 3159
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0680022671037376e-05,
+      "loss": 1.2721,
+      "step": 3160
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0667968296227169e-05,
+      "loss": 1.2578,
+      "step": 3161
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0655918253389452e-05,
+      "loss": 1.2648,
+      "step": 3162
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.06438725481179e-05,
+      "loss": 1.2056,
+      "step": 3163
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.06318311860042e-05,
+      "loss": 1.2825,
+      "step": 3164
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0619794172637995e-05,
+      "loss": 1.2496,
+      "step": 3165
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0607761513606935e-05,
+      "loss": 1.2619,
+      "step": 3166
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0595733214496633e-05,
+      "loss": 1.251,
+      "step": 3167
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0583709280890668e-05,
+      "loss": 1.2337,
+      "step": 3168
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0571689718370629e-05,
+      "loss": 1.2735,
+      "step": 3169
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0559674532516033e-05,
+      "loss": 1.2749,
+      "step": 3170
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0547663728904392e-05,
+      "loss": 1.2735,
+      "step": 3171
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0535657313111183e-05,
+      "loss": 1.2379,
+      "step": 3172
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0523655290709825e-05,
+      "loss": 1.2867,
+      "step": 3173
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0511657667271731e-05,
+      "loss": 1.2593,
+      "step": 3174
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0499664448366245e-05,
+      "loss": 1.246,
+      "step": 3175
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0487675639560664e-05,
+      "loss": 1.3067,
+      "step": 3176
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0475691246420267e-05,
+      "loss": 1.2627,
+      "step": 3177
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0463711274508253e-05,
+      "loss": 1.2061,
+      "step": 3178
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.045173572938579e-05,
+      "loss": 1.2407,
+      "step": 3179
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0439764616611972e-05,
+      "loss": 1.2155,
+      "step": 3180
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0427797941743854e-05,
+      "loss": 1.2505,
+      "step": 3181
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.041583571033641e-05,
+      "loss": 1.2293,
+      "step": 3182
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0403877927942582e-05,
+      "loss": 1.2287,
+      "step": 3183
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0391924600113211e-05,
+      "loss": 1.1778,
+      "step": 3184
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0379975732397096e-05,
+      "loss": 1.2483,
+      "step": 3185
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0368031330340948e-05,
+      "loss": 1.1827,
+      "step": 3186
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0356091399489431e-05,
+      "loss": 1.2479,
+      "step": 3187
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0344155945385106e-05,
+      "loss": 1.2303,
+      "step": 3188
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0332224973568458e-05,
+      "loss": 1.1771,
+      "step": 3189
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0320298489577913e-05,
+      "loss": 1.2381,
+      "step": 3190
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.03083764989498e-05,
+      "loss": 1.2529,
+      "step": 3191
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0296459007218345e-05,
+      "loss": 1.2194,
+      "step": 3192
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0284546019915727e-05,
+      "loss": 1.2178,
+      "step": 3193
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0272637542571988e-05,
+      "loss": 1.228,
+      "step": 3194
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.026073358071512e-05,
+      "loss": 1.2682,
+      "step": 3195
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0248834139870985e-05,
+      "loss": 1.2121,
+      "step": 3196
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0236939225563351e-05,
+      "loss": 1.2508,
+      "step": 3197
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0225048843313914e-05,
+      "loss": 1.216,
+      "step": 3198
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.021316299864223e-05,
+      "loss": 1.1624,
+      "step": 3199
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0201281697065757e-05,
+      "loss": 1.2211,
+      "step": 3200
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0189404944099867e-05,
+      "loss": 1.2744,
+      "step": 3201
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.017753274525779e-05,
+      "loss": 1.215,
+      "step": 3202
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.016566510605067e-05,
+      "loss": 1.217,
+      "step": 3203
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0153802031987504e-05,
+      "loss": 1.25,
+      "step": 3204
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0141943528575205e-05,
+      "loss": 1.2136,
+      "step": 3205
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0130089601318525e-05,
+      "loss": 1.2514,
+      "step": 3206
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0118240255720128e-05,
+      "loss": 1.2354,
+      "step": 3207
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0106395497280524e-05,
+      "loss": 1.2434,
+      "step": 3208
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0094555331498118e-05,
+      "loss": 1.2592,
+      "step": 3209
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.0082719763869153e-05,
+      "loss": 1.1914,
+      "step": 3210
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.0070888799887772e-05,
+      "loss": 1.2514,
+      "step": 3211
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.0059062445045957e-05,
+      "loss": 1.2839,
+      "step": 3212
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.0047240704833544e-05,
+      "loss": 1.2125,
+      "step": 3213
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.0035423584738262e-05,
+      "loss": 1.2671,
+      "step": 3214
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.0023611090245653e-05,
+      "loss": 1.2732,
+      "step": 3215
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.0011803226839148e-05,
+      "loss": 1.2836,
+      "step": 3216
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.0000000000000006e-05,
+      "loss": 1.2816,
+      "step": 3217
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.988201415207327e-06,
+      "loss": 1.2156,
+      "step": 3218
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.976407477938092e-06,
+      "loss": 1.2123,
+      "step": 3219
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.96461819366709e-06,
+      "loss": 1.2187,
+      "step": 3220
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.952833567866954e-06,
+      "loss": 1.257,
+      "step": 3221
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.941053606008176e-06,
+      "loss": 1.2064,
+      "step": 3222
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.929278313559054e-06,
+      "loss": 1.2562,
+      "step": 3223
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.917507695985752e-06,
+      "loss": 1.2479,
+      "step": 3224
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.905741758752234e-06,
+      "loss": 1.2899,
+      "step": 3225
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.893980507320295e-06,
+      "loss": 1.2659,
+      "step": 3226
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.882223947149583e-06,
+      "loss": 1.2383,
+      "step": 3227
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.870472083697526e-06,
+      "loss": 1.2468,
+      "step": 3228
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.858724922419413e-06,
+      "loss": 1.2514,
+      "step": 3229
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.846982468768316e-06,
+      "loss": 1.2496,
+      "step": 3230
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.83524472819515e-06,
+      "loss": 1.2408,
+      "step": 3231
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.823511706148612e-06,
+      "loss": 1.2441,
+      "step": 3232
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.811783408075244e-06,
+      "loss": 1.2472,
+      "step": 3233
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.800059839419358e-06,
+      "loss": 1.2464,
+      "step": 3234
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.788341005623107e-06,
+      "loss": 1.2239,
+      "step": 3235
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.776626912126413e-06,
+      "loss": 1.2192,
+      "step": 3236
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.764917564367025e-06,
+      "loss": 1.2329,
+      "step": 3237
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.753212967780472e-06,
+      "loss": 1.2609,
+      "step": 3238
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.741513127800072e-06,
+      "loss": 1.2723,
+      "step": 3239
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.729818049856963e-06,
+      "loss": 1.2593,
+      "step": 3240
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.718127739380043e-06,
+      "loss": 1.2197,
+      "step": 3241
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.706442201796007e-06,
+      "loss": 1.2573,
+      "step": 3242
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.694761442529345e-06,
+      "loss": 1.196,
+      "step": 3243
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.683085467002306e-06,
+      "loss": 1.2149,
+      "step": 3244
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.67141428063495e-06,
+      "loss": 1.2241,
+      "step": 3245
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.659747888845087e-06,
+      "loss": 1.2784,
+      "step": 3246
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.648086297048302e-06,
+      "loss": 1.2894,
+      "step": 3247
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.636429510657974e-06,
+      "loss": 1.2243,
+      "step": 3248
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.624777535085233e-06,
+      "loss": 1.1729,
+      "step": 3249
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.61313037573897e-06,
+      "loss": 1.2249,
+      "step": 3250
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.601488038025869e-06,
+      "loss": 1.2333,
+      "step": 3251
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.589850527350337e-06,
+      "loss": 1.2303,
+      "step": 3252
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.578217849114579e-06,
+      "loss": 1.2754,
+      "step": 3253
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.566590008718524e-06,
+      "loss": 1.2453,
+      "step": 3254
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.554967011559874e-06,
+      "loss": 1.1803,
+      "step": 3255
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.54334886303409e-06,
+      "loss": 1.2374,
+      "step": 3256
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.53173556853435e-06,
+      "loss": 1.2788,
+      "step": 3257
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.520127133451619e-06,
+      "loss": 1.1753,
+      "step": 3258
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.508523563174578e-06,
+      "loss": 1.2168,
+      "step": 3259
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.496924863089652e-06,
+      "loss": 1.2706,
+      "step": 3260
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.485331038581021e-06,
+      "loss": 1.1991,
+      "step": 3261
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.473742095030588e-06,
+      "loss": 1.2774,
+      "step": 3262
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.46215803781799e-06,
+      "loss": 1.256,
+      "step": 3263
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.450578872320613e-06,
+      "loss": 1.2213,
+      "step": 3264
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.439004603913542e-06,
+      "loss": 1.2804,
+      "step": 3265
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.427435237969624e-06,
+      "loss": 1.2548,
+      "step": 3266
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.415870779859405e-06,
+      "loss": 1.2665,
+      "step": 3267
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.404311234951148e-06,
+      "loss": 1.2561,
+      "step": 3268
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.392756608610871e-06,
+      "loss": 1.239,
+      "step": 3269
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.381206906202268e-06,
+      "loss": 1.2774,
+      "step": 3270
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.369662133086768e-06,
+      "loss": 1.2445,
+      "step": 3271
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.358122294623514e-06,
+      "loss": 1.2767,
+      "step": 3272
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.34658739616934e-06,
+      "loss": 1.2274,
+      "step": 3273
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.335057443078817e-06,
+      "loss": 1.2329,
+      "step": 3274
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.323532440704196e-06,
+      "loss": 1.2633,
+      "step": 3275
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.312012394395423e-06,
+      "loss": 1.2442,
+      "step": 3276
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.300497309500176e-06,
+      "loss": 1.2476,
+      "step": 3277
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.288987191363799e-06,
+      "loss": 1.2524,
+      "step": 3278
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.277482045329344e-06,
+      "loss": 1.2107,
+      "step": 3279
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.265981876737566e-06,
+      "loss": 1.235,
+      "step": 3280
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.254486690926878e-06,
+      "loss": 1.2543,
+      "step": 3281
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.242996493233414e-06,
+      "loss": 1.281,
+      "step": 3282
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.23151128899097e-06,
+      "loss": 1.2388,
+      "step": 3283
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.220031083531026e-06,
+      "loss": 1.2788,
+      "step": 3284
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.208555882182762e-06,
+      "loss": 1.2302,
+      "step": 3285
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.197085690273e-06,
+      "loss": 1.2492,
+      "step": 3286
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.185620513126275e-06,
+      "loss": 1.2273,
+      "step": 3287
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.174160356064765e-06,
+      "loss": 1.2494,
+      "step": 3288
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.162705224408326e-06,
+      "loss": 1.2193,
+      "step": 3289
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.151255123474493e-06,
+      "loss": 1.275,
+      "step": 3290
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.139810058578451e-06,
+      "loss": 1.2566,
+      "step": 3291
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.128370035033046e-06,
+      "loss": 1.2343,
+      "step": 3292
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.116935058148801e-06,
+      "loss": 1.2535,
+      "step": 3293
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.105505133233876e-06,
+      "loss": 1.2429,
+      "step": 3294
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.094080265594108e-06,
+      "loss": 1.2582,
+      "step": 3295
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.082660460532961e-06,
+      "loss": 1.2402,
+      "step": 3296
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.071245723351563e-06,
+      "loss": 1.2902,
+      "step": 3297
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.059836059348696e-06,
+      "loss": 1.2472,
+      "step": 3298
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.048431473820776e-06,
+      "loss": 1.2695,
+      "step": 3299
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.037031972061854e-06,
+      "loss": 1.2196,
+      "step": 3300
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.02563755936365e-06,
+      "loss": 1.2352,
+      "step": 3301
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.014248241015484e-06,
+      "loss": 1.2558,
+      "step": 3302
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.00286402230434e-06,
+      "loss": 1.2743,
+      "step": 3303
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.991484908514835e-06,
+      "loss": 1.2453,
+      "step": 3304
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.980110904929189e-06,
+      "loss": 1.2606,
+      "step": 3305
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.968742016827283e-06,
+      "loss": 1.2222,
+      "step": 3306
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.957378249486592e-06,
+      "loss": 1.2485,
+      "step": 3307
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.946019608182245e-06,
+      "loss": 1.3003,
+      "step": 3308
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.93466609818697e-06,
+      "loss": 1.2299,
+      "step": 3309
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.92331772477111e-06,
+      "loss": 1.2602,
+      "step": 3310
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.91197449320265e-06,
+      "loss": 1.2744,
+      "step": 3311
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.900636408747156e-06,
+      "loss": 1.2881,
+      "step": 3312
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.889303476667823e-06,
+      "loss": 1.2524,
+      "step": 3313
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.877975702225457e-06,
+      "loss": 1.2903,
+      "step": 3314
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.866653090678452e-06,
+      "loss": 1.257,
+      "step": 3315
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.855335647282833e-06,
+      "loss": 1.2713,
+      "step": 3316
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.844023377292198e-06,
+      "loss": 1.2491,
+      "step": 3317
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.832716285957754e-06,
+      "loss": 1.2851,
+      "step": 3318
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.821414378528314e-06,
+      "loss": 1.2564,
+      "step": 3319
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.810117660250275e-06,
+      "loss": 1.2711,
+      "step": 3320
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.798826136367613e-06,
+      "loss": 1.2451,
+      "step": 3321
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.787539812121924e-06,
+      "loss": 1.2593,
+      "step": 3322
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.776258692752355e-06,
+      "loss": 1.2359,
+      "step": 3323
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.76498278349567e-06,
+      "loss": 1.267,
+      "step": 3324
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.753712089586184e-06,
+      "loss": 1.2868,
+      "step": 3325
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.74244661625582e-06,
+      "loss": 1.2327,
+      "step": 3326
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.731186368734049e-06,
+      "loss": 1.2488,
+      "step": 3327
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.719931352247944e-06,
+      "loss": 1.2485,
+      "step": 3328
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.708681572022122e-06,
+      "loss": 1.2541,
+      "step": 3329
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.697437033278797e-06,
+      "loss": 1.2607,
+      "step": 3330
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.686197741237722e-06,
+      "loss": 1.2404,
+      "step": 3331
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.674963701116243e-06,
+      "loss": 1.2154,
+      "step": 3332
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.663734918129247e-06,
+      "loss": 1.2064,
+      "step": 3333
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.652511397489181e-06,
+      "loss": 1.2387,
+      "step": 3334
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.641293144406067e-06,
+      "loss": 1.2578,
+      "step": 3335
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.630080164087456e-06,
+      "loss": 1.2699,
+      "step": 3336
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.618872461738483e-06,
+      "loss": 1.212,
+      "step": 3337
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.607670042561807e-06,
+      "loss": 1.26,
+      "step": 3338
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.596472911757633e-06,
+      "loss": 1.2571,
+      "step": 3339
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.58528107452374e-06,
+      "loss": 1.2404,
+      "step": 3340
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.574094536055423e-06,
+      "loss": 1.261,
+      "step": 3341
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.562913301545513e-06,
+      "loss": 1.2811,
+      "step": 3342
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.55173737618441e-06,
+      "loss": 1.2974,
+      "step": 3343
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.540566765160016e-06,
+      "loss": 1.2205,
+      "step": 3344
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.529401473657795e-06,
+      "loss": 1.2413,
+      "step": 3345
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.518241506860719e-06,
+      "loss": 1.2487,
+      "step": 3346
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.507086869949287e-06,
+      "loss": 1.2469,
+      "step": 3347
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.495937568101551e-06,
+      "loss": 1.2834,
+      "step": 3348
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.484793606493054e-06,
+      "loss": 1.2322,
+      "step": 3349
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.473654990296887e-06,
+      "loss": 1.2866,
+      "step": 3350
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.462521724683637e-06,
+      "loss": 1.202,
+      "step": 3351
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.451393814821427e-06,
+      "loss": 1.2378,
+      "step": 3352
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.440271265875875e-06,
+      "loss": 1.2491,
+      "step": 3353
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.42915408301013e-06,
+      "loss": 1.214,
+      "step": 3354
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.418042271384828e-06,
+      "loss": 1.2479,
+      "step": 3355
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.406935836158138e-06,
+      "loss": 1.209,
+      "step": 3356
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.39583478248571e-06,
+      "loss": 1.235,
+      "step": 3357
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.3847391155207e-06,
+      "loss": 1.27,
+      "step": 3358
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.373648840413781e-06,
+      "loss": 1.2391,
+      "step": 3359
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.362563962313095e-06,
+      "loss": 1.2589,
+      "step": 3360
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.35148448636431e-06,
+      "loss": 1.1859,
+      "step": 3361
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.340410417710562e-06,
+      "loss": 1.2299,
+      "step": 3362
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.32934176149248e-06,
+      "loss": 1.2465,
+      "step": 3363
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.318278522848198e-06,
+      "loss": 1.2249,
+      "step": 3364
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.307220706913308e-06,
+      "loss": 1.2563,
+      "step": 3365
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.296168318820914e-06,
+      "loss": 1.2431,
+      "step": 3366
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.28512136370158e-06,
+      "loss": 1.2983,
+      "step": 3367
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.274079846683346e-06,
+      "loss": 1.257,
+      "step": 3368
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.263043772891752e-06,
+      "loss": 1.2664,
+      "step": 3369
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.252013147449785e-06,
+      "loss": 1.2095,
+      "step": 3370
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.240987975477903e-06,
+      "loss": 1.2258,
+      "step": 3371
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.229968262094064e-06,
+      "loss": 1.2169,
+      "step": 3372
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.218954012413647e-06,
+      "loss": 1.2153,
+      "step": 3373
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.207945231549539e-06,
+      "loss": 1.2195,
+      "step": 3374
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.19694192461205e-06,
+      "loss": 1.2949,
+      "step": 3375
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.185944096708982e-06,
+      "loss": 1.23,
+      "step": 3376
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.17495175294556e-06,
+      "loss": 1.2486,
+      "step": 3377
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.1639648984245e-06,
+      "loss": 1.2274,
+      "step": 3378
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.152983538245933e-06,
+      "loss": 1.2346,
+      "step": 3379
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.142007677507475e-06,
+      "loss": 1.284,
+      "step": 3380
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.131037321304154e-06,
+      "loss": 1.2561,
+      "step": 3381
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.120072474728476e-06,
+      "loss": 1.2514,
+      "step": 3382
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.10911314287037e-06,
+      "loss": 1.2614,
+      "step": 3383
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.098159330817192e-06,
+      "loss": 1.2331,
+      "step": 3384
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.087211043653777e-06,
+      "loss": 1.2503,
+      "step": 3385
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.076268286462352e-06,
+      "loss": 1.2748,
+      "step": 3386
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.065331064322612e-06,
+      "loss": 1.2431,
+      "step": 3387
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.054399382311657e-06,
+      "loss": 1.1891,
+      "step": 3388
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.043473245504017e-06,
+      "loss": 1.2109,
+      "step": 3389
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.032552658971672e-06,
+      "loss": 1.2667,
+      "step": 3390
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.021637627784e-06,
+      "loss": 1.2332,
+      "step": 3391
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.010728157007805e-06,
+      "loss": 1.183,
+      "step": 3392
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.999824251707324e-06,
+      "loss": 1.2525,
+      "step": 3393
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.98892591694419e-06,
+      "loss": 1.256,
+      "step": 3394
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.978033157777473e-06,
+      "loss": 1.2587,
+      "step": 3395
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.967145979263637e-06,
+      "loss": 1.2686,
+      "step": 3396
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.956264386456551e-06,
+      "loss": 1.2134,
+      "step": 3397
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.945388384407518e-06,
+      "loss": 1.244,
+      "step": 3398
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.934517978165211e-06,
+      "loss": 1.2408,
+      "step": 3399
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.92365317277574e-06,
+      "loss": 1.2673,
+      "step": 3400
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.912793973282584e-06,
+      "loss": 1.2307,
+      "step": 3401
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.90194038472665e-06,
+      "loss": 1.256,
+      "step": 3402
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.891092412146204e-06,
+      "loss": 1.2364,
+      "step": 3403
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.880250060576943e-06,
+      "loss": 1.2482,
+      "step": 3404
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.869413335051926e-06,
+      "loss": 1.2059,
+      "step": 3405
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.85858224060162e-06,
+      "loss": 1.2578,
+      "step": 3406
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.847756782253864e-06,
+      "loss": 1.1665,
+      "step": 3407
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.836936965033881e-06,
+      "loss": 1.2442,
+      "step": 3408
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.826122793964293e-06,
+      "loss": 1.2477,
+      "step": 3409
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.815314274065074e-06,
+      "loss": 1.2205,
+      "step": 3410
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.804511410353603e-06,
+      "loss": 1.2614,
+      "step": 3411
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.793714207844616e-06,
+      "loss": 1.2646,
+      "step": 3412
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.782922671550213e-06,
+      "loss": 1.2035,
+      "step": 3413
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.772136806479891e-06,
+      "loss": 1.1502,
+      "step": 3414
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.761356617640485e-06,
+      "loss": 1.2283,
+      "step": 3415
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.750582110036225e-06,
+      "loss": 1.2111,
+      "step": 3416
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.739813288668677e-06,
+      "loss": 1.257,
+      "step": 3417
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.72905015853677e-06,
+      "loss": 1.2329,
+      "step": 3418
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.718292724636815e-06,
+      "loss": 1.2531,
+      "step": 3419
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.70754099196246e-06,
+      "loss": 1.2305,
+      "step": 3420
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.696794965504695e-06,
+      "loss": 1.2241,
+      "step": 3421
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.686054650251893e-06,
+      "loss": 1.2656,
+      "step": 3422
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.675320051189746e-06,
+      "loss": 1.1752,
+      "step": 3423
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.664591173301315e-06,
+      "loss": 1.2414,
+      "step": 3424
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.65386802156698e-06,
+      "loss": 1.2726,
+      "step": 3425
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.64315060096449e-06,
+      "loss": 1.2527,
+      "step": 3426
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.632438916468928e-06,
+      "loss": 1.2549,
+      "step": 3427
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.621732973052696e-06,
+      "loss": 1.1833,
+      "step": 3428
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.611032775685541e-06,
+      "loss": 1.2045,
+      "step": 3429
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.600338329334554e-06,
+      "loss": 1.2253,
+      "step": 3430
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.5896496389641336e-06,
+      "loss": 1.2404,
+      "step": 3431
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.5789667095360355e-06,
+      "loss": 1.2634,
+      "step": 3432
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.568289546009316e-06,
+      "loss": 1.2626,
+      "step": 3433
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.557618153340358e-06,
+      "loss": 1.2234,
+      "step": 3434
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.546952536482888e-06,
+      "loss": 1.2331,
+      "step": 3435
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.536292700387924e-06,
+      "loss": 1.2757,
+      "step": 3436
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.5256386500038055e-06,
+      "loss": 1.1817,
+      "step": 3437
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.5149903902762066e-06,
+      "loss": 1.2456,
+      "step": 3438
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.504347926148086e-06,
+      "loss": 1.2407,
+      "step": 3439
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.4937112625597375e-06,
+      "loss": 1.2644,
+      "step": 3440
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.483080404448744e-06,
+      "loss": 1.2944,
+      "step": 3441
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.472455356749992e-06,
+      "loss": 1.2336,
+      "step": 3442
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.461836124395692e-06,
+      "loss": 1.2524,
+      "step": 3443
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.451222712315325e-06,
+      "loss": 1.2425,
+      "step": 3444
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.440615125435702e-06,
+      "loss": 1.2094,
+      "step": 3445
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.430013368680908e-06,
+      "loss": 1.2546,
+      "step": 3446
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.419417446972319e-06,
+      "loss": 1.255,
+      "step": 3447
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.408827365228625e-06,
+      "loss": 1.1557,
+      "step": 3448
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.3982431283657805e-06,
+      "loss": 1.2077,
+      "step": 3449
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.38766474129704e-06,
+      "loss": 1.2892,
+      "step": 3450
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.37709220893295e-06,
+      "loss": 1.2841,
+      "step": 3451
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.3665255361813125e-06,
+      "loss": 1.2524,
+      "step": 3452
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.355964727947242e-06,
+      "loss": 1.2555,
+      "step": 3453
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.3454097891331085e-06,
+      "loss": 1.2726,
+      "step": 3454
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.334860724638555e-06,
+      "loss": 1.244,
+      "step": 3455
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.3243175393605215e-06,
+      "loss": 1.2741,
+      "step": 3456
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.313780238193195e-06,
+      "loss": 1.2557,
+      "step": 3457
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.303248826028036e-06,
+      "loss": 1.1983,
+      "step": 3458
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.292723307753784e-06,
+      "loss": 1.2621,
+      "step": 3459
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.282203688256422e-06,
+      "loss": 1.2212,
+      "step": 3460
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.27168997241922e-06,
+      "loss": 1.2561,
+      "step": 3461
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.261182165122689e-06,
+      "loss": 1.27,
+      "step": 3462
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.250680271244593e-06,
+      "loss": 1.2527,
+      "step": 3463
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.240184295659971e-06,
+      "loss": 1.2466,
+      "step": 3464
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.229694243241097e-06,
+      "loss": 1.2089,
+      "step": 3465
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.219210118857509e-06,
+      "loss": 1.2399,
+      "step": 3466
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.208731927375982e-06,
+      "loss": 1.2491,
+      "step": 3467
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.198259673660535e-06,
+      "loss": 1.2486,
+      "step": 3468
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.187793362572451e-06,
+      "loss": 1.2738,
+      "step": 3469
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.17733299897023e-06,
+      "loss": 1.2042,
+      "step": 3470
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.166878587709618e-06,
+      "loss": 1.2367,
+      "step": 3471
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.156430133643613e-06,
+      "loss": 1.257,
+      "step": 3472
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.145987641622423e-06,
+      "loss": 1.2147,
+      "step": 3473
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.1355511164935085e-06,
+      "loss": 1.202,
+      "step": 3474
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.125120563101562e-06,
+      "loss": 1.2162,
+      "step": 3475
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.114695986288476e-06,
+      "loss": 1.2245,
+      "step": 3476
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.104277390893404e-06,
+      "loss": 1.2367,
+      "step": 3477
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.0938647817527014e-06,
+      "loss": 1.235,
+      "step": 3478
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.083458163699939e-06,
+      "loss": 1.2113,
+      "step": 3479
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.073057541565933e-06,
+      "loss": 1.2623,
+      "step": 3480
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.062662920178689e-06,
+      "loss": 1.2144,
+      "step": 3481
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.052274304363449e-06,
+      "loss": 1.2208,
+      "step": 3482
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.041891698942649e-06,
+      "loss": 1.2248,
+      "step": 3483
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.03151510873594e-06,
+      "loss": 1.2647,
+      "step": 3484
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.021144538560194e-06,
+      "loss": 1.2,
+      "step": 3485
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.010779993229471e-06,
+      "loss": 1.2185,
+      "step": 3486
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.000421477555038e-06,
+      "loss": 1.2572,
+      "step": 3487
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.9900689963453734e-06,
+      "loss": 1.2395,
+      "step": 3488
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.9797225544061385e-06,
+      "loss": 1.2401,
+      "step": 3489
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.969382156540212e-06,
+      "loss": 1.2825,
+      "step": 3490
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.9590478075476475e-06,
+      "loss": 1.2534,
+      "step": 3491
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.9487195122256925e-06,
+      "loss": 1.2292,
+      "step": 3492
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.9383972753688e-06,
+      "loss": 1.2419,
+      "step": 3493
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.928081101768589e-06,
+      "loss": 1.2715,
+      "step": 3494
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.9177709962138905e-06,
+      "loss": 1.2377,
+      "step": 3495
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.907466963490692e-06,
+      "loss": 1.2476,
+      "step": 3496
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.897169008382172e-06,
+      "loss": 1.2539,
+      "step": 3497
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.88687713566869e-06,
+      "loss": 1.2817,
+      "step": 3498
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.876591350127795e-06,
+      "loss": 1.2159,
+      "step": 3499
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.866311656534177e-06,
+      "loss": 1.2417,
+      "step": 3500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.856038059659731e-06,
+      "loss": 1.2626,
+      "step": 3501
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.8457705642734994e-06,
+      "loss": 1.2404,
+      "step": 3502
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.835509175141713e-06,
+      "loss": 1.2234,
+      "step": 3503
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.825253897027746e-06,
+      "loss": 1.2569,
+      "step": 3504
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.815004734692146e-06,
+      "loss": 1.2727,
+      "step": 3505
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.804761692892627e-06,
+      "loss": 1.2193,
+      "step": 3506
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.794524776384059e-06,
+      "loss": 1.2128,
+      "step": 3507
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.784293989918454e-06,
+      "loss": 1.2532,
+      "step": 3508
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.774069338245002e-06,
+      "loss": 1.2278,
+      "step": 3509
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.763850826110025e-06,
+      "loss": 1.2912,
+      "step": 3510
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.753638458257017e-06,
+      "loss": 1.2289,
+      "step": 3511
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.743432239426599e-06,
+      "loss": 1.2061,
+      "step": 3512
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.733232174356537e-06,
+      "loss": 1.182,
+      "step": 3513
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.723038267781763e-06,
+      "loss": 1.2255,
+      "step": 3514
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.712850524434329e-06,
+      "loss": 1.2829,
+      "step": 3515
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.7026689490434275e-06,
+      "loss": 1.2193,
+      "step": 3516
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.692493546335404e-06,
+      "loss": 1.2078,
+      "step": 3517
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.682324321033715e-06,
+      "loss": 1.215,
+      "step": 3518
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.672161277858977e-06,
+      "loss": 1.2736,
+      "step": 3519
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.662004421528909e-06,
+      "loss": 1.1534,
+      "step": 3520
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.651853756758382e-06,
+      "loss": 1.2636,
+      "step": 3521
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.641709288259368e-06,
+      "loss": 1.2854,
+      "step": 3522
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.6315710207409925e-06,
+      "loss": 1.2688,
+      "step": 3523
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.621438958909472e-06,
+      "loss": 1.2235,
+      "step": 3524
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.6113131074681694e-06,
+      "loss": 1.2121,
+      "step": 3525
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.6011934711175395e-06,
+      "loss": 1.2806,
+      "step": 3526
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.591080054555177e-06,
+      "loss": 1.2419,
+      "step": 3527
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.580972862475769e-06,
+      "loss": 1.2342,
+      "step": 3528
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.570871899571119e-06,
+      "loss": 1.2547,
+      "step": 3529
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.56077717053015e-06,
+      "loss": 1.214,
+      "step": 3530
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.550688680038871e-06,
+      "loss": 1.2362,
+      "step": 3531
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.5406064327804165e-06,
+      "loss": 1.2625,
+      "step": 3532
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.5305304334350075e-06,
+      "loss": 1.2325,
+      "step": 3533
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.520460686679964e-06,
+      "loss": 1.2307,
+      "step": 3534
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.510397197189724e-06,
+      "loss": 1.2151,
+      "step": 3535
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.500339969635794e-06,
+      "loss": 1.1559,
+      "step": 3536
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.490289008686786e-06,
+      "loss": 1.2368,
+      "step": 3537
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.480244319008411e-06,
+      "loss": 1.2878,
+      "step": 3538
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.470205905263449e-06,
+      "loss": 1.2264,
+      "step": 3539
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.460173772111791e-06,
+      "loss": 1.2734,
+      "step": 3540
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.450147924210395e-06,
+      "loss": 1.1985,
+      "step": 3541
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.440128366213297e-06,
+      "loss": 1.227,
+      "step": 3542
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.430115102771637e-06,
+      "loss": 1.2165,
+      "step": 3543
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.420108138533607e-06,
+      "loss": 1.2112,
+      "step": 3544
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.410107478144496e-06,
+      "loss": 1.2751,
+      "step": 3545
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.400113126246645e-06,
+      "loss": 1.2595,
+      "step": 3546
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.390125087479493e-06,
+      "loss": 1.2717,
+      "step": 3547
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.380143366479521e-06,
+      "loss": 1.249,
+      "step": 3548
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.370167967880303e-06,
+      "loss": 1.1961,
+      "step": 3549
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.360198896312451e-06,
+      "loss": 1.2486,
+      "step": 3550
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.350236156403666e-06,
+      "loss": 1.2977,
+      "step": 3551
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.3402797527786904e-06,
+      "loss": 1.1871,
+      "step": 3552
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.330329690059342e-06,
+      "loss": 1.3087,
+      "step": 3553
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.32038597286448e-06,
+      "loss": 1.201,
+      "step": 3554
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.31044860581002e-06,
+      "loss": 1.248,
+      "step": 3555
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.300517593508944e-06,
+      "loss": 1.1977,
+      "step": 3556
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.290592940571269e-06,
+      "loss": 1.2783,
+      "step": 3557
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.280674651604059e-06,
+      "loss": 1.2428,
+      "step": 3558
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.270762731211442e-06,
+      "loss": 1.2188,
+      "step": 3559
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.260857183994564e-06,
+      "loss": 1.2497,
+      "step": 3560
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.25095801455164e-06,
+      "loss": 1.2437,
+      "step": 3561
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.241065227477905e-06,
+      "loss": 1.2495,
+      "step": 3562
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.231178827365627e-06,
+      "loss": 1.2344,
+      "step": 3563
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.221298818804136e-06,
+      "loss": 1.2486,
+      "step": 3564
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.211425206379769e-06,
+      "loss": 1.225,
+      "step": 3565
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.201557994675895e-06,
+      "loss": 1.2596,
+      "step": 3566
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.191697188272933e-06,
+      "loss": 1.2276,
+      "step": 3567
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.181842791748307e-06,
+      "loss": 1.2601,
+      "step": 3568
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.17199480967648e-06,
+      "loss": 1.2197,
+      "step": 3569
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.162153246628921e-06,
+      "loss": 1.2276,
+      "step": 3570
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.152318107174144e-06,
+      "loss": 1.2539,
+      "step": 3571
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.142489395877651e-06,
+      "loss": 1.2383,
+      "step": 3572
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.132667117301989e-06,
+      "loss": 1.2276,
+      "step": 3573
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.122851276006692e-06,
+      "loss": 1.2338,
+      "step": 3574
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.113041876548333e-06,
+      "loss": 1.2444,
+      "step": 3575
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.103238923480468e-06,
+      "loss": 1.2727,
+      "step": 3576
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.093442421353683e-06,
+      "loss": 1.2161,
+      "step": 3577
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.083652374715561e-06,
+      "loss": 1.2587,
+      "step": 3578
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.073868788110673e-06,
+      "loss": 1.2926,
+      "step": 3579
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.064091666080621e-06,
+      "loss": 1.2441,
+      "step": 3580
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.054321013163978e-06,
+      "loss": 1.2681,
+      "step": 3581
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.044556833896338e-06,
+      "loss": 1.2571,
+      "step": 3582
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.034799132810274e-06,
+      "loss": 1.2116,
+      "step": 3583
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.025047914435349e-06,
+      "loss": 1.2729,
+      "step": 3584
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.015303183298135e-06,
+      "loss": 1.2212,
+      "step": 3585
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.005564943922179e-06,
+      "loss": 1.2731,
+      "step": 3586
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.995833200828007e-06,
+      "loss": 1.2619,
+      "step": 3587
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.9861079585331535e-06,
+      "loss": 1.2284,
+      "step": 3588
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.97638922155211e-06,
+      "loss": 1.2559,
+      "step": 3589
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.9666769943963675e-06,
+      "loss": 1.3126,
+      "step": 3590
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.956971281574386e-06,
+      "loss": 1.2676,
+      "step": 3591
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.947272087591591e-06,
+      "loss": 1.2489,
+      "step": 3592
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.9375794169504095e-06,
+      "loss": 1.2458,
+      "step": 3593
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.927893274150214e-06,
+      "loss": 1.192,
+      "step": 3594
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.918213663687362e-06,
+      "loss": 1.2844,
+      "step": 3595
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.908540590055168e-06,
+      "loss": 1.2122,
+      "step": 3596
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.898874057743926e-06,
+      "loss": 1.2162,
+      "step": 3597
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.889214071240876e-06,
+      "loss": 1.2739,
+      "step": 3598
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.879560635030242e-06,
+      "loss": 1.3043,
+      "step": 3599
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.869913753593175e-06,
+      "loss": 1.2569,
+      "step": 3600
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.860273431407821e-06,
+      "loss": 1.1974,
+      "step": 3601
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.8506396729492455e-06,
+      "loss": 1.236,
+      "step": 3602
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.841012482689501e-06,
+      "loss": 1.2349,
+      "step": 3603
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.831391865097564e-06,
+      "loss": 1.191,
+      "step": 3604
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.821777824639365e-06,
+      "loss": 1.2687,
+      "step": 3605
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.812170365777801e-06,
+      "loss": 1.2637,
+      "step": 3606
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.802569492972687e-06,
+      "loss": 1.2613,
+      "step": 3607
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.792975210680793e-06,
+      "loss": 1.2483,
+      "step": 3608
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.783387523355839e-06,
+      "loss": 1.2237,
+      "step": 3609
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.773806435448459e-06,
+      "loss": 1.2205,
+      "step": 3610
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.764231951406256e-06,
+      "loss": 1.2307,
+      "step": 3611
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.75466407567374e-06,
+      "loss": 1.2022,
+      "step": 3612
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.745102812692358e-06,
+      "loss": 1.194,
+      "step": 3613
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.735548166900506e-06,
+      "loss": 1.232,
+      "step": 3614
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.72600014273349e-06,
+      "loss": 1.2266,
+      "step": 3615
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.716458744623536e-06,
+      "loss": 1.223,
+      "step": 3616
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.706923976999825e-06,
+      "loss": 1.2352,
+      "step": 3617
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.697395844288423e-06,
+      "loss": 1.2387,
+      "step": 3618
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.687874350912346e-06,
+      "loss": 1.2801,
+      "step": 3619
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.678359501291504e-06,
+      "loss": 1.2382,
+      "step": 3620
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.668851299842739e-06,
+      "loss": 1.2551,
+      "step": 3621
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.659349750979814e-06,
+      "loss": 1.2399,
+      "step": 3622
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.6498548591133725e-06,
+      "loss": 1.2322,
+      "step": 3623
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.6403666286510065e-06,
+      "loss": 1.2039,
+      "step": 3624
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.630885063997187e-06,
+      "loss": 1.2529,
+      "step": 3625
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.621410169553292e-06,
+      "loss": 1.2361,
+      "step": 3626
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.6119419497176275e-06,
+      "loss": 1.192,
+      "step": 3627
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.6024804088853775e-06,
+      "loss": 1.2415,
+      "step": 3628
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.59302555144863e-06,
+      "loss": 1.2426,
+      "step": 3629
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.5835773817963814e-06,
+      "loss": 1.2766,
+      "step": 3630
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.574135904314504e-06,
+      "loss": 1.253,
+      "step": 3631
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.5647011233857915e-06,
+      "loss": 1.205,
+      "step": 3632
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.555273043389906e-06,
+      "loss": 1.2142,
+      "step": 3633
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.545851668703397e-06,
+      "loss": 1.2438,
+      "step": 3634
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.536437003699724e-06,
+      "loss": 1.2575,
+      "step": 3635
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.527029052749216e-06,
+      "loss": 1.1944,
+      "step": 3636
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.517627820219076e-06,
+      "loss": 1.2532,
+      "step": 3637
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.508233310473412e-06,
+      "loss": 1.2462,
+      "step": 3638
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.498845527873193e-06,
+      "loss": 1.2677,
+      "step": 3639
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.489464476776276e-06,
+      "loss": 1.2286,
+      "step": 3640
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.480090161537388e-06,
+      "loss": 1.196,
+      "step": 3641
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.470722586508122e-06,
+      "loss": 1.2695,
+      "step": 3642
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.4613617560369625e-06,
+      "loss": 1.2407,
+      "step": 3643
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.452007674469235e-06,
+      "loss": 1.2691,
+      "step": 3644
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.442660346147157e-06,
+      "loss": 1.2453,
+      "step": 3645
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.433319775409807e-06,
+      "loss": 1.2642,
+      "step": 3646
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.4239859665931105e-06,
+      "loss": 1.254,
+      "step": 3647
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.4146589240298745e-06,
+      "loss": 1.2766,
+      "step": 3648
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.405338652049749e-06,
+      "loss": 1.2345,
+      "step": 3649
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.396025154979247e-06,
+      "loss": 1.179,
+      "step": 3650
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.386718437141743e-06,
+      "loss": 1.1928,
+      "step": 3651
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.37741850285745e-06,
+      "loss": 1.2355,
+      "step": 3652
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.368125356443452e-06,
+      "loss": 1.2487,
+      "step": 3653
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.358839002213665e-06,
+      "loss": 1.2252,
+      "step": 3654
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.349559444478849e-06,
+      "loss": 1.2632,
+      "step": 3655
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.3402866875466344e-06,
+      "loss": 1.2128,
+      "step": 3656
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.331020735721469e-06,
+      "loss": 1.2661,
+      "step": 3657
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.321761593304646e-06,
+      "loss": 1.2443,
+      "step": 3658
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.312509264594312e-06,
+      "loss": 1.1673,
+      "step": 3659
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.303263753885433e-06,
+      "loss": 1.2504,
+      "step": 3660
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.294025065469827e-06,
+      "loss": 1.2509,
+      "step": 3661
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.284793203636132e-06,
+      "loss": 1.3018,
+      "step": 3662
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.2755681726698134e-06,
+      "loss": 1.2068,
+      "step": 3663
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.2663499768531865e-06,
+      "loss": 1.2447,
+      "step": 3664
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.257138620465374e-06,
+      "loss": 1.1832,
+      "step": 3665
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.247934107782324e-06,
+      "loss": 1.2177,
+      "step": 3666
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.238736443076828e-06,
+      "loss": 1.2891,
+      "step": 3667
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.2295456306184715e-06,
+      "loss": 1.2142,
+      "step": 3668
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.220361674673677e-06,
+      "loss": 1.2576,
+      "step": 3669
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.211184579505688e-06,
+      "loss": 1.2489,
+      "step": 3670
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.2020143493745425e-06,
+      "loss": 1.2028,
+      "step": 3671
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.192850988537115e-06,
+      "loss": 1.2788,
+      "step": 3672
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.183694501247072e-06,
+      "loss": 1.2287,
+      "step": 3673
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.174544891754896e-06,
+      "loss": 1.2406,
+      "step": 3674
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.165402164307884e-06,
+      "loss": 1.2041,
+      "step": 3675
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.15626632315013e-06,
+      "loss": 1.257,
+      "step": 3676
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.147137372522537e-06,
+      "loss": 1.2025,
+      "step": 3677
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.138015316662803e-06,
+      "loss": 1.2163,
+      "step": 3678
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.128900159805425e-06,
+      "loss": 1.2846,
+      "step": 3679
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.119791906181713e-06,
+      "loss": 1.2558,
+      "step": 3680
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.110690560019744e-06,
+      "loss": 1.2866,
+      "step": 3681
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.1015961255444235e-06,
+      "loss": 1.2177,
+      "step": 3682
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.09250860697742e-06,
+      "loss": 1.2401,
+      "step": 3683
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.083428008537197e-06,
+      "loss": 1.259,
+      "step": 3684
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.074354334439022e-06,
+      "loss": 1.2806,
+      "step": 3685
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.065287588894933e-06,
+      "loss": 1.1889,
+      "step": 3686
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.056227776113747e-06,
+      "loss": 1.2237,
+      "step": 3687
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.0471749003010835e-06,
+      "loss": 1.2335,
+      "step": 3688
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.038128965659317e-06,
+      "loss": 1.2801,
+      "step": 3689
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.029089976387627e-06,
+      "loss": 1.2576,
+      "step": 3690
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.020057936681939e-06,
+      "loss": 1.2338,
+      "step": 3691
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.011032850734983e-06,
+      "loss": 1.2521,
+      "step": 3692
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.00201472273623e-06,
+      "loss": 1.2056,
+      "step": 3693
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.993003556871954e-06,
+      "loss": 1.2287,
+      "step": 3694
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.983999357325164e-06,
+      "loss": 1.1885,
+      "step": 3695
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.975002128275666e-06,
+      "loss": 1.2171,
+      "step": 3696
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.966011873900001e-06,
+      "loss": 1.2356,
+      "step": 3697
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.957028598371498e-06,
+      "loss": 1.2419,
+      "step": 3698
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.948052305860233e-06,
+      "loss": 1.2049,
+      "step": 3699
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.939083000533036e-06,
+      "loss": 1.2135,
+      "step": 3700
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.93012068655351e-06,
+      "loss": 1.1876,
+      "step": 3701
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.92116536808199e-06,
+      "loss": 1.2686,
+      "step": 3702
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.912217049275594e-06,
+      "loss": 1.2107,
+      "step": 3703
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.903275734288162e-06,
+      "loss": 1.2027,
+      "step": 3704
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.8943414272702886e-06,
+      "loss": 1.2154,
+      "step": 3705
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.885414132369335e-06,
+      "loss": 1.2347,
+      "step": 3706
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.876493853729385e-06,
+      "loss": 1.218,
+      "step": 3707
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.867580595491268e-06,
+      "loss": 1.2118,
+      "step": 3708
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.858674361792571e-06,
+      "loss": 1.2323,
+      "step": 3709
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.849775156767598e-06,
+      "loss": 1.2283,
+      "step": 3710
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.840882984547415e-06,
+      "loss": 1.2607,
+      "step": 3711
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.8319978492598e-06,
+      "loss": 1.2752,
+      "step": 3712
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.823119755029271e-06,
+      "loss": 1.2268,
+      "step": 3713
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.814248705977092e-06,
+      "loss": 1.2078,
+      "step": 3714
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.805384706221232e-06,
+      "loss": 1.2239,
+      "step": 3715
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.796527759876415e-06,
+      "loss": 1.199,
+      "step": 3716
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.787677871054062e-06,
+      "loss": 1.2697,
+      "step": 3717
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7788350438623465e-06,
+      "loss": 1.2864,
+      "step": 3718
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.769999282406137e-06,
+      "loss": 1.2448,
+      "step": 3719
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7611705907870474e-06,
+      "loss": 1.2648,
+      "step": 3720
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7523489731033845e-06,
+      "loss": 1.2404,
+      "step": 3721
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.743534433450199e-06,
+      "loss": 1.2325,
+      "step": 3722
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.734726975919233e-06,
+      "loss": 1.2223,
+      "step": 3723
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.725926604598942e-06,
+      "loss": 1.2159,
+      "step": 3724
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7171333235745145e-06,
+      "loss": 1.24,
+      "step": 3725
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.708347136927818e-06,
+      "loss": 1.2243,
+      "step": 3726
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.699568048737453e-06,
+      "loss": 1.2913,
+      "step": 3727
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.690796063078709e-06,
+      "loss": 1.1972,
+      "step": 3728
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.6820311840235745e-06,
+      "loss": 1.2271,
+      "step": 3729
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.67327341564076e-06,
+      "loss": 1.2028,
+      "step": 3730
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.6645227619956515e-06,
+      "loss": 1.2191,
+      "step": 3731
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.655779227150352e-06,
+      "loss": 1.2429,
+      "step": 3732
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.647042815163649e-06,
+      "loss": 1.1935,
+      "step": 3733
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.638313530091016e-06,
+      "loss": 1.2538,
+      "step": 3734
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.629591375984641e-06,
+      "loss": 1.2513,
+      "step": 3735
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.620876356893385e-06,
+      "loss": 1.204,
+      "step": 3736
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.612168476862789e-06,
+      "loss": 1.2585,
+      "step": 3737
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.603467739935108e-06,
+      "loss": 1.2116,
+      "step": 3738
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.594774150149251e-06,
+      "loss": 1.2384,
+      "step": 3739
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.586087711540832e-06,
+      "loss": 1.2116,
+      "step": 3740
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.57740842814213e-06,
+      "loss": 1.2345,
+      "step": 3741
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.568736303982115e-06,
+      "loss": 1.2151,
+      "step": 3742
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.560071343086421e-06,
+      "loss": 1.2314,
+      "step": 3743
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.55141354947737e-06,
+      "loss": 1.2186,
+      "step": 3744
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.542762927173941e-06,
+      "loss": 1.2214,
+      "step": 3745
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.534119480191801e-06,
+      "loss": 1.1697,
+      "step": 3746
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.525483212543273e-06,
+      "loss": 1.2306,
+      "step": 3747
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.516854128237358e-06,
+      "loss": 1.2209,
+      "step": 3748
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.5082322312797166e-06,
+      "loss": 1.259,
+      "step": 3749
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.499617525672664e-06,
+      "loss": 1.22,
+      "step": 3750
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.491010015415198e-06,
+      "loss": 1.2254,
+      "step": 3751
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.4824097045029615e-06,
+      "loss": 1.1593,
+      "step": 3752
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.473816596928251e-06,
+      "loss": 1.2386,
+      "step": 3753
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.465230696680038e-06,
+      "loss": 1.1959,
+      "step": 3754
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.456652007743929e-06,
+      "loss": 1.2521,
+      "step": 3755
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.448080534102202e-06,
+      "loss": 1.2475,
+      "step": 3756
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.439516279733764e-06,
+      "loss": 1.2411,
+      "step": 3757
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.430959248614184e-06,
+      "loss": 1.2321,
+      "step": 3758
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.42240944471568e-06,
+      "loss": 1.2352,
+      "step": 3759
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.413866872007104e-06,
+      "loss": 1.1946,
+      "step": 3760
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.405331534453967e-06,
+      "loss": 1.2099,
+      "step": 3761
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.396803436018406e-06,
+      "loss": 1.2542,
+      "step": 3762
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.3882825806592024e-06,
+      "loss": 1.1668,
+      "step": 3763
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.379768972331784e-06,
+      "loss": 1.3,
+      "step": 3764
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.371262614988196e-06,
+      "loss": 1.275,
+      "step": 3765
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.362763512577144e-06,
+      "loss": 1.2259,
+      "step": 3766
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.354271669043934e-06,
+      "loss": 1.1798,
+      "step": 3767
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.345787088330537e-06,
+      "loss": 1.2543,
+      "step": 3768
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.33730977437552e-06,
+      "loss": 1.2457,
+      "step": 3769
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.328839731114101e-06,
+      "loss": 1.2836,
+      "step": 3770
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.3203769624781055e-06,
+      "loss": 1.2497,
+      "step": 3771
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.311921472395999e-06,
+      "loss": 1.2518,
+      "step": 3772
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.303473264792857e-06,
+      "loss": 1.2356,
+      "step": 3773
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.295032343590366e-06,
+      "loss": 1.2162,
+      "step": 3774
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.286598712706858e-06,
+      "loss": 1.243,
+      "step": 3775
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.278172376057246e-06,
+      "loss": 1.2294,
+      "step": 3776
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.269753337553091e-06,
+      "loss": 1.2745,
+      "step": 3777
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2613416011025424e-06,
+      "loss": 1.2469,
+      "step": 3778
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.25293717061036e-06,
+      "loss": 1.2469,
+      "step": 3779
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.244540049977934e-06,
+      "loss": 1.2199,
+      "step": 3780
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.236150243103234e-06,
+      "loss": 1.2239,
+      "step": 3781
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.227767753880861e-06,
+      "loss": 1.2538,
+      "step": 3782
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.2193925862019934e-06,
+      "loss": 1.2101,
+      "step": 3783
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.211024743954424e-06,
+      "loss": 1.2115,
+      "step": 3784
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.2026642310225505e-06,
+      "loss": 1.2154,
+      "step": 3785
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.194311051287359e-06,
+      "loss": 1.2042,
+      "step": 3786
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.185965208626428e-06,
+      "loss": 1.2834,
+      "step": 3787
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.177626706913948e-06,
+      "loss": 1.2421,
+      "step": 3788
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.16929555002068e-06,
+      "loss": 1.2578,
+      "step": 3789
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.160971741813995e-06,
+      "loss": 1.1894,
+      "step": 3790
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.152655286157834e-06,
+      "loss": 1.2586,
+      "step": 3791
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.144346186912738e-06,
+      "loss": 1.2576,
+      "step": 3792
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.136044447935837e-06,
+      "loss": 1.3064,
+      "step": 3793
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.127750073080829e-06,
+      "loss": 1.2095,
+      "step": 3794
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.119463066197997e-06,
+      "loss": 1.2668,
+      "step": 3795
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.111183431134223e-06,
+      "loss": 1.2512,
+      "step": 3796
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.102911171732933e-06,
+      "loss": 1.2132,
+      "step": 3797
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.094646291834166e-06,
+      "loss": 1.2339,
+      "step": 3798
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.086388795274508e-06,
+      "loss": 1.2086,
+      "step": 3799
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.078138685887125e-06,
+      "loss": 1.2069,
+      "step": 3800
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.069895967501765e-06,
+      "loss": 1.1852,
+      "step": 3801
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.0616606439447315e-06,
+      "loss": 1.2256,
+      "step": 3802
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.053432719038895e-06,
+      "loss": 1.2168,
+      "step": 3803
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.045212196603705e-06,
+      "loss": 1.2169,
+      "step": 3804
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.03699908045516e-06,
+      "loss": 1.2411,
+      "step": 3805
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.028793374405833e-06,
+      "loss": 1.2408,
+      "step": 3806
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.020595082264847e-06,
+      "loss": 1.2576,
+      "step": 3807
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.012404207837881e-06,
+      "loss": 1.2201,
+      "step": 3808
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.0042207549271905e-06,
+      "loss": 1.2566,
+      "step": 3809
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.996044727331558e-06,
+      "loss": 1.2368,
+      "step": 3810
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.987876128846349e-06,
+      "loss": 1.2535,
+      "step": 3811
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.979714963263455e-06,
+      "loss": 1.2385,
+      "step": 3812
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.971561234371324e-06,
+      "loss": 1.2306,
+      "step": 3813
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.963414945954962e-06,
+      "loss": 1.2366,
+      "step": 3814
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.955276101795908e-06,
+      "loss": 1.2534,
+      "step": 3815
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.947144705672257e-06,
+      "loss": 1.2208,
+      "step": 3816
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.939020761358641e-06,
+      "loss": 1.241,
+      "step": 3817
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.930904272626226e-06,
+      "loss": 1.2647,
+      "step": 3818
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.922795243242734e-06,
+      "loss": 1.2261,
+      "step": 3819
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.914693676972408e-06,
+      "loss": 1.2204,
+      "step": 3820
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.906599577576027e-06,
+      "loss": 1.1952,
+      "step": 3821
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.898512948810922e-06,
+      "loss": 1.2485,
+      "step": 3822
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.890433794430934e-06,
+      "loss": 1.2584,
+      "step": 3823
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.882362118186445e-06,
+      "loss": 1.2613,
+      "step": 3824
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.87429792382437e-06,
+      "loss": 1.2003,
+      "step": 3825
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.86624121508814e-06,
+      "loss": 1.246,
+      "step": 3826
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.858191995717722e-06,
+      "loss": 1.2902,
+      "step": 3827
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.850150269449597e-06,
+      "loss": 1.2454,
+      "step": 3828
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.84211604001677e-06,
+      "loss": 1.2498,
+      "step": 3829
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.834089311148774e-06,
+      "loss": 1.2363,
+      "step": 3830
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.826070086571651e-06,
+      "loss": 1.2316,
+      "step": 3831
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.818058370007956e-06,
+      "loss": 1.2111,
+      "step": 3832
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.810054165176775e-06,
+      "loss": 1.2284,
+      "step": 3833
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.802057475793688e-06,
+      "loss": 1.1595,
+      "step": 3834
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.794068305570804e-06,
+      "loss": 1.2422,
+      "step": 3835
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.78608665821673e-06,
+      "loss": 1.2094,
+      "step": 3836
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.778112537436578e-06,
+      "loss": 1.2318,
+      "step": 3837
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7701459469319824e-06,
+      "loss": 1.268,
+      "step": 3838
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7621868904010585e-06,
+      "loss": 1.2058,
+      "step": 3839
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7542353715384462e-06,
+      "loss": 1.2071,
+      "step": 3840
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7462913940352797e-06,
+      "loss": 1.2856,
+      "step": 3841
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7383549615791826e-06,
+      "loss": 1.275,
+      "step": 3842
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7304260778542924e-06,
+      "loss": 1.2445,
+      "step": 3843
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.722504746541229e-06,
+      "loss": 1.2283,
+      "step": 3844
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.714590971317107e-06,
+      "loss": 1.2413,
+      "step": 3845
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.706684755855545e-06,
+      "loss": 1.2882,
+      "step": 3846
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.698786103826639e-06,
+      "loss": 1.2922,
+      "step": 3847
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.690895018896987e-06,
+      "loss": 1.2429,
+      "step": 3848
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6830115047296633e-06,
+      "loss": 1.2325,
+      "step": 3849
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.675135564984227e-06,
+      "loss": 1.1791,
+      "step": 3850
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6672672033167333e-06,
+      "loss": 1.2524,
+      "step": 3851
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6594064233797123e-06,
+      "loss": 1.2261,
+      "step": 3852
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6515532288221646e-06,
+      "loss": 1.2731,
+      "step": 3853
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.643707623289592e-06,
+      "loss": 1.2579,
+      "step": 3854
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.635869610423952e-06,
+      "loss": 1.2219,
+      "step": 3855
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.628039193863695e-06,
+      "loss": 1.2137,
+      "step": 3856
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6202163772437326e-06,
+      "loss": 1.2421,
+      "step": 3857
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6124011641954473e-06,
+      "loss": 1.272,
+      "step": 3858
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6045935583467053e-06,
+      "loss": 1.2398,
+      "step": 3859
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.5967935633218277e-06,
+      "loss": 1.2371,
+      "step": 3860
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.589001182741616e-06,
+      "loss": 1.2357,
+      "step": 3861
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.5812164202233236e-06,
+      "loss": 1.1327,
+      "step": 3862
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.5734392793806704e-06,
+      "loss": 1.1843,
+      "step": 3863
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.5656697638238447e-06,
+      "loss": 1.2136,
+      "step": 3864
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.5579078771594988e-06,
+      "loss": 1.2392,
+      "step": 3865
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.550153622990724e-06,
+      "loss": 1.2362,
+      "step": 3866
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.542407004917092e-06,
+      "loss": 1.1976,
+      "step": 3867
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.5346680265346113e-06,
+      "loss": 1.2348,
+      "step": 3868
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.5269366914357585e-06,
+      "loss": 1.2678,
+      "step": 3869
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.5192130032094517e-06,
+      "loss": 1.256,
+      "step": 3870
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.511496965441057e-06,
+      "loss": 1.1927,
+      "step": 3871
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.503788581712406e-06,
+      "loss": 1.2388,
+      "step": 3872
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.4960878556017597e-06,
+      "loss": 1.2157,
+      "step": 3873
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.488394790683829e-06,
+      "loss": 1.238,
+      "step": 3874
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.480709390529777e-06,
+      "loss": 1.2533,
+      "step": 3875
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.473031658707193e-06,
+      "loss": 1.1969,
+      "step": 3876
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.465361598780128e-06,
+      "loss": 1.2303,
+      "step": 3877
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.4576992143090517e-06,
+      "loss": 1.2805,
+      "step": 3878
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.450044508850876e-06,
+      "loss": 1.26,
+      "step": 3879
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.4423974859589594e-06,
+      "loss": 1.1821,
+      "step": 3880
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.4347581491830796e-06,
+      "loss": 1.2238,
+      "step": 3881
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.427126502069449e-06,
+      "loss": 1.2715,
+      "step": 3882
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.4195025481607224e-06,
+      "loss": 1.2255,
+      "step": 3883
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.411886290995965e-06,
+      "loss": 1.2507,
+      "step": 3884
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.4042777341106903e-06,
+      "loss": 1.2484,
+      "step": 3885
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3966768810368132e-06,
+      "loss": 1.2795,
+      "step": 3886
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3890837353026964e-06,
+      "loss": 1.2657,
+      "step": 3887
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3814983004331014e-06,
+      "loss": 1.2332,
+      "step": 3888
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.373920579949237e-06,
+      "loss": 1.2309,
+      "step": 3889
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3663505773687023e-06,
+      "loss": 1.2391,
+      "step": 3890
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3587882962055374e-06,
+      "loss": 1.2462,
+      "step": 3891
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3512337399701813e-06,
+      "loss": 1.2391,
+      "step": 3892
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3436869121695013e-06,
+      "loss": 1.2377,
+      "step": 3893
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3361478163067673e-06,
+      "loss": 1.2021,
+      "step": 3894
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.328616455881657e-06,
+      "loss": 1.2061,
+      "step": 3895
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3210928343902716e-06,
+      "loss": 1.2055,
+      "step": 3896
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3135769553251017e-06,
+      "loss": 1.2479,
+      "step": 3897
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3060688221750637e-06,
+      "loss": 1.2068,
+      "step": 3898
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.2985684384254648e-06,
+      "loss": 1.2059,
+      "step": 3899
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.2910758075580085e-06,
+      "loss": 1.2377,
+      "step": 3900
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.283590933050822e-06,
+      "loss": 1.2323,
+      "step": 3901
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.2761138183784126e-06,
+      "loss": 1.2227,
+      "step": 3902
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.2686444670116878e-06,
+      "loss": 1.1798,
+      "step": 3903
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.261182882417966e-06,
+      "loss": 1.2491,
+      "step": 3904
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.253729068060938e-06,
+      "loss": 1.2497,
+      "step": 3905
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.2462830274007073e-06,
+      "loss": 1.2374,
+      "step": 3906
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.2388447638937583e-06,
+      "loss": 1.237,
+      "step": 3907
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.2314142809929617e-06,
+      "loss": 1.2184,
+      "step": 3908
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.223991582147592e-06,
+      "loss": 1.2358,
+      "step": 3909
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.216576670803291e-06,
+      "loss": 1.1994,
+      "step": 3910
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.2091695504021047e-06,
+      "loss": 1.2259,
+      "step": 3911
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.2017702243824434e-06,
+      "loss": 1.2491,
+      "step": 3912
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.1943786961791166e-06,
+      "loss": 1.25,
+      "step": 3913
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.1869949692232982e-06,
+      "loss": 1.1986,
+      "step": 3914
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.179619046942557e-06,
+      "loss": 1.2398,
+      "step": 3915
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.172250932760823e-06,
+      "loss": 1.1818,
+      "step": 3916
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.164890630098416e-06,
+      "loss": 1.2342,
+      "step": 3917
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.1575381423720142e-06,
+      "loss": 1.2302,
+      "step": 3918
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.150193472994687e-06,
+      "loss": 1.2234,
+      "step": 3919
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.142856625375856e-06,
+      "loss": 1.2781,
+      "step": 3920
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.13552760292132e-06,
+      "loss": 1.2061,
+      "step": 3921
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.1282064090332522e-06,
+      "loss": 1.1835,
+      "step": 3922
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.1208930471101786e-06,
+      "loss": 1.219,
+      "step": 3923
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.1135875205469946e-06,
+      "loss": 1.2042,
+      "step": 3924
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.1062898327349656e-06,
+      "loss": 1.2249,
+      "step": 3925
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.098999987061706e-06,
+      "loss": 1.2185,
+      "step": 3926
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0917179869112023e-06,
+      "loss": 1.2449,
+      "step": 3927
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.084443835663791e-06,
+      "loss": 1.1974,
+      "step": 3928
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.077177536696159e-06,
+      "loss": 1.259,
+      "step": 3929
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0699190933813683e-06,
+      "loss": 1.2247,
+      "step": 3930
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0626685090888177e-06,
+      "loss": 1.2188,
+      "step": 3931
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0554257871842543e-06,
+      "loss": 1.2518,
+      "step": 3932
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0481909310297954e-06,
+      "loss": 1.2321,
+      "step": 3933
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0409639439838833e-06,
+      "loss": 1.2839,
+      "step": 3934
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0337448294013307e-06,
+      "loss": 1.2261,
+      "step": 3935
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0265335906332717e-06,
+      "loss": 1.1707,
+      "step": 3936
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.019330231027209e-06,
+      "loss": 1.2104,
+      "step": 3937
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.012134753926965e-06,
+      "loss": 1.2262,
+      "step": 3938
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0049471626727246e-06,
+      "loss": 1.2489,
+      "step": 3939
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.997767460600991e-06,
+      "loss": 1.2649,
+      "step": 3940
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.990595651044621e-06,
+      "loss": 1.1955,
+      "step": 3941
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9834317373327983e-06,
+      "loss": 1.2643,
+      "step": 3942
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.976275722791051e-06,
+      "loss": 1.2107,
+      "step": 3943
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9691276107412293e-06,
+      "loss": 1.2201,
+      "step": 3944
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.961987404501516e-06,
+      "loss": 1.2721,
+      "step": 3945
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9548551073864386e-06,
+      "loss": 1.2075,
+      "step": 3946
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.947730722706832e-06,
+      "loss": 1.236,
+      "step": 3947
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.94061425376988e-06,
+      "loss": 1.2498,
+      "step": 3948
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9335057038790715e-06,
+      "loss": 1.1947,
+      "step": 3949
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9264050763342267e-06,
+      "loss": 1.2651,
+      "step": 3950
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9193123744315e-06,
+      "loss": 1.2381,
+      "step": 3951
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.912227601463351e-06,
+      "loss": 1.2728,
+      "step": 3952
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9051507607185603e-06,
+      "loss": 1.2692,
+      "step": 3953
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8980818554822376e-06,
+      "loss": 1.2416,
+      "step": 3954
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8910208890357916e-06,
+      "loss": 1.2354,
+      "step": 3955
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.883967864656969e-06,
+      "loss": 1.2996,
+      "step": 3956
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.876922785619809e-06,
+      "loss": 1.2389,
+      "step": 3957
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8698856551946664e-06,
+      "loss": 1.1833,
+      "step": 3958
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8628564766482193e-06,
+      "loss": 1.2702,
+      "step": 3959
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.855835253243433e-06,
+      "loss": 1.2147,
+      "step": 3960
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.848821988239605e-06,
+      "loss": 1.242,
+      "step": 3961
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8418166848923158e-06,
+      "loss": 1.1824,
+      "step": 3962
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.834819346453468e-06,
+      "loss": 1.2445,
+      "step": 3963
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.827829976171248e-06,
+      "loss": 1.2354,
+      "step": 3964
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.820848577290165e-06,
+      "loss": 1.2483,
+      "step": 3965
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8138751530510065e-06,
+      "loss": 1.2257,
+      "step": 3966
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.806909706690881e-06,
+      "loss": 1.2184,
+      "step": 3967
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.799952241443167e-06,
+      "loss": 1.3169,
+      "step": 3968
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.7930027605375644e-06,
+      "loss": 1.2348,
+      "step": 3969
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7860612672000485e-06,
+      "loss": 1.2304,
+      "step": 3970
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7791277646528893e-06,
+      "loss": 1.2483,
+      "step": 3971
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.77220225611466e-06,
+      "loss": 1.2624,
+      "step": 3972
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7652847448002074e-06,
+      "loss": 1.224,
+      "step": 3973
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7583752339206714e-06,
+      "loss": 1.2417,
+      "step": 3974
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7514737266834845e-06,
+      "loss": 1.1722,
+      "step": 3975
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7445802262923505e-06,
+      "loss": 1.1919,
+      "step": 3976
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.737694735947276e-06,
+      "loss": 1.1951,
+      "step": 3977
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.730817258844529e-06,
+      "loss": 1.2739,
+      "step": 3978
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.723947798176665e-06,
+      "loss": 1.1915,
+      "step": 3979
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7170863571325257e-06,
+      "loss": 1.2676,
+      "step": 3980
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7102329388972215e-06,
+      "loss": 1.2477,
+      "step": 3981
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7033875466521363e-06,
+      "loss": 1.2373,
+      "step": 3982
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.696550183574942e-06,
+      "loss": 1.213,
+      "step": 3983
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.6897208528395656e-06,
+      "loss": 1.1821,
+      "step": 3984
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.682899557616223e-06,
+      "loss": 1.2195,
+      "step": 3985
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.676086301071381e-06,
+      "loss": 1.2169,
+      "step": 3986
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.66928108636779e-06,
+      "loss": 1.2245,
+      "step": 3987
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.662483916664467e-06,
+      "loss": 1.2297,
+      "step": 3988
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.6556947951166836e-06,
+      "loss": 1.1888,
+      "step": 3989
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.648913724875981e-06,
+      "loss": 1.2278,
+      "step": 3990
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.6421407090901707e-06,
+      "loss": 1.2259,
+      "step": 3991
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.635375750903306e-06,
+      "loss": 1.2322,
+      "step": 3992
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.628618853455727e-06,
+      "loss": 1.2886,
+      "step": 3993
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.621870019884005e-06,
+      "loss": 1.2436,
+      "step": 3994
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.6151292533209826e-06,
+      "loss": 1.2344,
+      "step": 3995
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.6083965568957603e-06,
+      "loss": 1.2718,
+      "step": 3996
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.601671933733678e-06,
+      "loss": 1.2734,
+      "step": 3997
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.594955386956346e-06,
+      "loss": 1.2458,
+      "step": 3998
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.588246919681614e-06,
+      "loss": 1.2649,
+      "step": 3999
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5815465350235756e-06,
+      "loss": 1.2268,
+      "step": 4000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5748542360925944e-06,
+      "loss": 1.2762,
+      "step": 4001
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.568170025995258e-06,
+      "loss": 1.2597,
+      "step": 4002
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.561493907834405e-06,
+      "loss": 1.2471,
+      "step": 4003
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5548258847091266e-06,
+      "loss": 1.2999,
+      "step": 4004
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.548165959714748e-06,
+      "loss": 1.1962,
+      "step": 4005
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.54151413594284e-06,
+      "loss": 1.2221,
+      "step": 4006
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.534870416481208e-06,
+      "loss": 1.2091,
+      "step": 4007
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5282348044138915e-06,
+      "loss": 1.1968,
+      "step": 4008
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.521607302821183e-06,
+      "loss": 1.2388,
+      "step": 4009
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.514987914779592e-06,
+      "loss": 1.1939,
+      "step": 4010
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5083766433618695e-06,
+      "loss": 1.2559,
+      "step": 4011
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5017734916370073e-06,
+      "loss": 1.2481,
+      "step": 4012
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.495178462670207e-06,
+      "loss": 1.2342,
+      "step": 4013
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.4885915595229215e-06,
+      "loss": 1.2446,
+      "step": 4014
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.4820127852528163e-06,
+      "loss": 1.2516,
+      "step": 4015
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.4754421429137887e-06,
+      "loss": 1.138,
+      "step": 4016
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.468879635555965e-06,
+      "loss": 1.231,
+      "step": 4017
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.462325266225687e-06,
+      "loss": 1.1968,
+      "step": 4018
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.455779037965529e-06,
+      "loss": 1.2174,
+      "step": 4019
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.4492409538142803e-06,
+      "loss": 1.263,
+      "step": 4020
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.44271101680694e-06,
+      "loss": 1.2434,
+      "step": 4021
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.436189229974748e-06,
+      "loss": 1.2122,
+      "step": 4022
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.4296755963451424e-06,
+      "loss": 1.2643,
+      "step": 4023
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.423170118941778e-06,
+      "loss": 1.2076,
+      "step": 4024
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.4166728007845364e-06,
+      "loss": 1.212,
+      "step": 4025
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.4101836448894924e-06,
+      "loss": 1.2976,
+      "step": 4026
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.4037026542689555e-06,
+      "loss": 1.2679,
+      "step": 4027
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3972298319314224e-06,
+      "loss": 1.2742,
+      "step": 4028
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3907651808816067e-06,
+      "loss": 1.2112,
+      "step": 4029
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.384308704120435e-06,
+      "loss": 1.2603,
+      "step": 4030
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3778604046450313e-06,
+      "loss": 1.2175,
+      "step": 4031
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.371420285448722e-06,
+      "loss": 1.1484,
+      "step": 4032
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.364988349521049e-06,
+      "loss": 1.2499,
+      "step": 4033
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.358564599847737e-06,
+      "loss": 1.2726,
+      "step": 4034
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.352149039410727e-06,
+      "loss": 1.207,
+      "step": 4035
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.345741671188153e-06,
+      "loss": 1.231,
+      "step": 4036
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.33934249815434e-06,
+      "loss": 1.2326,
+      "step": 4037
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3329515232798207e-06,
+      "loss": 1.259,
+      "step": 4038
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3265687495313106e-06,
+      "loss": 1.2424,
+      "step": 4039
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3201941798717176e-06,
+      "loss": 1.2398,
+      "step": 4040
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.313827817260159e-06,
+      "loss": 1.2136,
+      "step": 4041
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.307469664651918e-06,
+      "loss": 1.1884,
+      "step": 4042
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3011197249984886e-06,
+      "loss": 1.2505,
+      "step": 4043
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2947780012475396e-06,
+      "loss": 1.2538,
+      "step": 4044
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2884444963429188e-06,
+      "loss": 1.1803,
+      "step": 4045
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.282119213224683e-06,
+      "loss": 1.1689,
+      "step": 4046
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2758021548290478e-06,
+      "loss": 1.2234,
+      "step": 4047
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2694933240884277e-06,
+      "loss": 1.2584,
+      "step": 4048
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.263192723931409e-06,
+      "loss": 1.1257,
+      "step": 4049
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2569003572827543e-06,
+      "loss": 1.2695,
+      "step": 4050
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.250616227063418e-06,
+      "loss": 1.2412,
+      "step": 4051
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.244340336190518e-06,
+      "loss": 1.2503,
+      "step": 4052
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2380726875773507e-06,
+      "loss": 1.2265,
+      "step": 4053
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2318132841333906e-06,
+      "loss": 1.2291,
+      "step": 4054
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2255621287642805e-06,
+      "loss": 1.2869,
+      "step": 4055
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2193192243718385e-06,
+      "loss": 1.2072,
+      "step": 4056
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2130845738540475e-06,
+      "loss": 1.2413,
+      "step": 4057
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2068581801050557e-06,
+      "loss": 1.2431,
+      "step": 4058
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2006400460151923e-06,
+      "loss": 1.2413,
+      "step": 4059
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.1944301744709428e-06,
+      "loss": 1.2783,
+      "step": 4060
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.1882285683549555e-06,
+      "loss": 1.1931,
+      "step": 4061
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.1820352305460492e-06,
+      "loss": 1.2397,
+      "step": 4062
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.1758501639191908e-06,
+      "loss": 1.1991,
+      "step": 4063
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.169673371345531e-06,
+      "loss": 1.2471,
+      "step": 4064
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.1635048556923555e-06,
+      "loss": 1.2106,
+      "step": 4065
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.1573446198231185e-06,
+      "loss": 1.1624,
+      "step": 4066
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.1511926665974324e-06,
+      "loss": 1.2528,
+      "step": 4067
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.1450489988710644e-06,
+      "loss": 1.1853,
+      "step": 4068
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.138913619495928e-06,
+      "loss": 1.1842,
+      "step": 4069
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.1327865313201015e-06,
+      "loss": 1.2136,
+      "step": 4070
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.1266677371877996e-06,
+      "loss": 1.1817,
+      "step": 4071
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.120557239939405e-06,
+      "loss": 1.1964,
+      "step": 4072
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.114455042411432e-06,
+      "loss": 1.2304,
+      "step": 4073
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.108361147436546e-06,
+      "loss": 1.2244,
+      "step": 4074
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.1022755578435715e-06,
+      "loss": 1.2338,
+      "step": 4075
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0961982764574597e-06,
+      "loss": 1.1968,
+      "step": 4076
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0901293060993154e-06,
+      "loss": 1.2226,
+      "step": 4077
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0840686495863837e-06,
+      "loss": 1.2201,
+      "step": 4078
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.078016309732047e-06,
+      "loss": 1.2608,
+      "step": 4079
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0719722893458317e-06,
+      "loss": 1.2443,
+      "step": 4080
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0659365912333972e-06,
+      "loss": 1.202,
+      "step": 4081
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0599092181965474e-06,
+      "loss": 1.1398,
+      "step": 4082
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0538901730332128e-06,
+      "loss": 1.2773,
+      "step": 4083
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.047879458537465e-06,
+      "loss": 1.2794,
+      "step": 4084
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0418770774995034e-06,
+      "loss": 1.2047,
+      "step": 4085
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0358830327056633e-06,
+      "loss": 1.2397,
+      "step": 4086
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0298973269384037e-06,
+      "loss": 1.2516,
+      "step": 4087
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.023919962976324e-06,
+      "loss": 1.2235,
+      "step": 4088
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0179509435941403e-06,
+      "loss": 1.1876,
+      "step": 4089
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.011990271562696e-06,
+      "loss": 1.2351,
+      "step": 4090
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.006037949648971e-06,
+      "loss": 1.2192,
+      "step": 4091
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.000093980616051e-06,
+      "loss": 1.2409,
+      "step": 4092
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9941583672231624e-06,
+      "loss": 1.2603,
+      "step": 4093
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9882311122256425e-06,
+      "loss": 1.2002,
+      "step": 4094
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9823122183749443e-06,
+      "loss": 1.2564,
+      "step": 4095
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9764016884186545e-06,
+      "loss": 1.1684,
+      "step": 4096
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9704995251004622e-06,
+      "loss": 1.2472,
+      "step": 4097
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9646057311601853e-06,
+      "loss": 1.2206,
+      "step": 4098
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.958720309333746e-06,
+      "loss": 1.1955,
+      "step": 4099
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.952843262353181e-06,
+      "loss": 1.1952,
+      "step": 4100
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.946974592946651e-06,
+      "loss": 1.212,
+      "step": 4101
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9411143038384163e-06,
+      "loss": 1.1871,
+      "step": 4102
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.935262397748845e-06,
+      "loss": 1.2422,
+      "step": 4103
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.929418877394429e-06,
+      "loss": 1.1622,
+      "step": 4104
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.923583745487747e-06,
+      "loss": 1.2082,
+      "step": 4105
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.917757004737506e-06,
+      "loss": 1.2459,
+      "step": 4106
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9119386578484934e-06,
+      "loss": 1.2342,
+      "step": 4107
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.906128707521624e-06,
+      "loss": 1.2289,
+      "step": 4108
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.900327156453896e-06,
+      "loss": 1.2199,
+      "step": 4109
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.894534007338422e-06,
+      "loss": 1.221,
+      "step": 4110
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8887492628644022e-06,
+      "loss": 1.1752,
+      "step": 4111
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8829729257171503e-06,
+      "loss": 1.2397,
+      "step": 4112
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8772049985780616e-06,
+      "loss": 1.216,
+      "step": 4113
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.871445484124641e-06,
+      "loss": 1.2169,
+      "step": 4114
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8656943850304765e-06,
+      "loss": 1.2489,
+      "step": 4115
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8599517039652548e-06,
+      "loss": 1.2228,
+      "step": 4116
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8542174435947614e-06,
+      "loss": 1.2122,
+      "step": 4117
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8484916065808622e-06,
+      "loss": 1.2304,
+      "step": 4118
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8427741955815138e-06,
+      "loss": 1.2397,
+      "step": 4119
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8370652132507705e-06,
+      "loss": 1.1906,
+      "step": 4120
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8313646622387639e-06,
+      "loss": 1.2633,
+      "step": 4121
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8256725451917233e-06,
+      "loss": 1.2508,
+      "step": 4122
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8199888647519537e-06,
+      "loss": 1.1996,
+      "step": 4123
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8143136235578374e-06,
+      "loss": 1.2186,
+      "step": 4124
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8086468242438582e-06,
+      "loss": 1.1936,
+      "step": 4125
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8029884694405631e-06,
+      "loss": 1.3133,
+      "step": 4126
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7973385617745953e-06,
+      "loss": 1.1846,
+      "step": 4127
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7916971038686614e-06,
+      "loss": 1.2062,
+      "step": 4128
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7860640983415533e-06,
+      "loss": 1.2634,
+      "step": 4129
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7804395478081416e-06,
+      "loss": 1.2399,
+      "step": 4130
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.774823454879362e-06,
+      "loss": 1.2037,
+      "step": 4131
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7692158221622379e-06,
+      "loss": 1.1931,
+      "step": 4132
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.763616652259854e-06,
+      "loss": 1.2231,
+      "step": 4133
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.758025947771378e-06,
+      "loss": 1.2224,
+      "step": 4134
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.752443711292029e-06,
+      "loss": 1.1991,
+      "step": 4135
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7468699454131211e-06,
+      "loss": 1.2411,
+      "step": 4136
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.741304652722009e-06,
+      "loss": 1.2054,
+      "step": 4137
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7357478358021374e-06,
+      "loss": 1.2458,
+      "step": 4138
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7301994972330028e-06,
+      "loss": 1.2315,
+      "step": 4139
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.724659639590167e-06,
+      "loss": 1.2182,
+      "step": 4140
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7191282654452646e-06,
+      "loss": 1.1663,
+      "step": 4141
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7136053773659766e-06,
+      "loss": 1.2193,
+      "step": 4142
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7080909779160615e-06,
+      "loss": 1.269,
+      "step": 4143
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7025850696553248e-06,
+      "loss": 1.2466,
+      "step": 4144
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6970876551396309e-06,
+      "loss": 1.2437,
+      "step": 4145
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6915987369209142e-06,
+      "loss": 1.2028,
+      "step": 4146
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6861183175471495e-06,
+      "loss": 1.2666,
+      "step": 4147
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6806463995623735e-06,
+      "loss": 1.2602,
+      "step": 4148
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6751829855066804e-06,
+      "loss": 1.1981,
+      "step": 4149
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.669728077916206e-06,
+      "loss": 1.2195,
+      "step": 4150
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6642816793231499e-06,
+      "loss": 1.2094,
+      "step": 4151
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6588437922557533e-06,
+      "loss": 1.2068,
+      "step": 4152
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6534144192383038e-06,
+      "loss": 1.2506,
+      "step": 4153
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6479935627911481e-06,
+      "loss": 1.2311,
+      "step": 4154
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6425812254306707e-06,
+      "loss": 1.2384,
+      "step": 4155
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.637177409669304e-06,
+      "loss": 1.2,
+      "step": 4156
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6317821180155214e-06,
+      "loss": 1.2337,
+      "step": 4157
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6263953529738464e-06,
+      "loss": 1.2668,
+      "step": 4158
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.621017117044843e-06,
+      "loss": 1.2193,
+      "step": 4159
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6156474127251077e-06,
+      "loss": 1.2489,
+      "step": 4160
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6102862425072818e-06,
+      "loss": 1.2418,
+      "step": 4161
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6049336088800505e-06,
+      "loss": 1.215,
+      "step": 4162
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5995895143281236e-06,
+      "loss": 1.2294,
+      "step": 4163
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5942539613322638e-06,
+      "loss": 1.1854,
+      "step": 4164
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5889269523692541e-06,
+      "loss": 1.205,
+      "step": 4165
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5836084899119165e-06,
+      "loss": 1.1918,
+      "step": 4166
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5782985764291091e-06,
+      "loss": 1.2247,
+      "step": 4167
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5729972143857164e-06,
+      "loss": 1.2258,
+      "step": 4168
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.567704406242654e-06,
+      "loss": 1.1779,
+      "step": 4169
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5624201544568717e-06,
+      "loss": 1.2531,
+      "step": 4170
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.557144461481337e-06,
+      "loss": 1.1995,
+      "step": 4171
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5518773297650613e-06,
+      "loss": 1.2137,
+      "step": 4172
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5466187617530647e-06,
+      "loss": 1.2479,
+      "step": 4173
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.541368759886397e-06,
+      "loss": 1.0979,
+      "step": 4174
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5361273266021392e-06,
+      "loss": 1.2153,
+      "step": 4175
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5308944643333857e-06,
+      "loss": 1.2625,
+      "step": 4176
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5256701755092574e-06,
+      "loss": 1.2282,
+      "step": 4177
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5204544625548922e-06,
+      "loss": 1.2628,
+      "step": 4178
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5152473278914447e-06,
+      "loss": 1.2377,
+      "step": 4179
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5100487739360993e-06,
+      "loss": 1.2581,
+      "step": 4180
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5048588031020405e-06,
+      "loss": 1.228,
+      "step": 4181
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4996774177984818e-06,
+      "loss": 1.2054,
+      "step": 4182
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.494504620430648e-06,
+      "loss": 1.25,
+      "step": 4183
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.489340413399769e-06,
+      "loss": 1.2274,
+      "step": 4184
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.484184799103101e-06,
+      "loss": 1.2219,
+      "step": 4185
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4790377799339007e-06,
+      "loss": 1.2518,
+      "step": 4186
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4738993582814343e-06,
+      "loss": 1.2306,
+      "step": 4187
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4687695365309895e-06,
+      "loss": 1.2787,
+      "step": 4188
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4636483170638505e-06,
+      "loss": 1.2192,
+      "step": 4189
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4585357022573043e-06,
+      "loss": 1.2671,
+      "step": 4190
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4534316944846595e-06,
+      "loss": 1.2599,
+      "step": 4191
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4483362961152114e-06,
+      "loss": 1.1853,
+      "step": 4192
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4432495095142796e-06,
+      "loss": 1.2101,
+      "step": 4193
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.438171337043164e-06,
+      "loss": 1.2596,
+      "step": 4194
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4331017810591764e-06,
+      "loss": 1.2582,
+      "step": 4195
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4280408439156369e-06,
+      "loss": 1.2308,
+      "step": 4196
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4229885279618461e-06,
+      "loss": 1.2049,
+      "step": 4197
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4179448355431168e-06,
+      "loss": 1.2557,
+      "step": 4198
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4129097690007543e-06,
+      "loss": 1.2506,
+      "step": 4199
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4078833306720573e-06,
+      "loss": 1.216,
+      "step": 4200
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4028655228903286e-06,
+      "loss": 1.2186,
+      "step": 4201
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.3978563479848538e-06,
+      "loss": 1.2294,
+      "step": 4202
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.3928558082809107e-06,
+      "loss": 1.2112,
+      "step": 4203
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.3878639060997822e-06,
+      "loss": 1.2523,
+      "step": 4204
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.3828806437587216e-06,
+      "loss": 1.3306,
+      "step": 4205
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.3779060235709918e-06,
+      "loss": 1.2433,
+      "step": 4206
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.3729400478458322e-06,
+      "loss": 1.1959,
+      "step": 4207
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3679827188884675e-06,
+      "loss": 1.2551,
+      "step": 4208
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3630340390001195e-06,
+      "loss": 1.1318,
+      "step": 4209
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.358094010477986e-06,
+      "loss": 1.2404,
+      "step": 4210
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.353162635615246e-06,
+      "loss": 1.2242,
+      "step": 4211
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3482399167010752e-06,
+      "loss": 1.2531,
+      "step": 4212
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3433258560206165e-06,
+      "loss": 1.2312,
+      "step": 4213
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3384204558550028e-06,
+      "loss": 1.2286,
+      "step": 4214
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.333523718481342e-06,
+      "loss": 1.1752,
+      "step": 4215
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3286356461727202e-06,
+      "loss": 1.1919,
+      "step": 4216
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3237562411982086e-06,
+      "loss": 1.2456,
+      "step": 4217
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3188855058228468e-06,
+      "loss": 1.2068,
+      "step": 4218
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3140234423076504e-06,
+      "loss": 1.2077,
+      "step": 4219
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3091700529096186e-06,
+      "loss": 1.2678,
+      "step": 4220
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.304325339881709e-06,
+      "loss": 1.2467,
+      "step": 4221
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2994893054728653e-06,
+      "loss": 1.2708,
+      "step": 4222
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2946619519279979e-06,
+      "loss": 1.2972,
+      "step": 4223
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2898432814879813e-06,
+      "loss": 1.2105,
+      "step": 4224
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2850332963896706e-06,
+      "loss": 1.2809,
+      "step": 4225
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2802319988658818e-06,
+      "loss": 1.2574,
+      "step": 4226
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2754393911453944e-06,
+      "loss": 1.2115,
+      "step": 4227
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2706554754529665e-06,
+      "loss": 1.2213,
+      "step": 4228
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2658802540093084e-06,
+      "loss": 1.2547,
+      "step": 4229
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2611137290311003e-06,
+      "loss": 1.2278,
+      "step": 4230
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2563559027309925e-06,
+      "loss": 1.1846,
+      "step": 4231
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2516067773175822e-06,
+      "loss": 1.2716,
+      "step": 4232
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2468663549954397e-06,
+      "loss": 1.2317,
+      "step": 4233
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2421346379650868e-06,
+      "loss": 1.2159,
+      "step": 4234
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2374116284230153e-06,
+      "loss": 1.2525,
+      "step": 4235
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2326973285616628e-06,
+      "loss": 1.2559,
+      "step": 4236
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2279917405694298e-06,
+      "loss": 1.239,
+      "step": 4237
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2232948666306732e-06,
+      "loss": 1.2,
+      "step": 4238
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.218606708925707e-06,
+      "loss": 1.2679,
+      "step": 4239
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2139272696307857e-06,
+      "loss": 1.2491,
+      "step": 4240
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2092565509181386e-06,
+      "loss": 1.2268,
+      "step": 4241
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2045945549559269e-06,
+      "loss": 1.2602,
+      "step": 4242
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1999412839082748e-06,
+      "loss": 1.1739,
+      "step": 4243
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1952967399352522e-06,
+      "loss": 1.232,
+      "step": 4244
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1906609251928746e-06,
+      "loss": 1.2015,
+      "step": 4245
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.186033841833112e-06,
+      "loss": 1.2532,
+      "step": 4246
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1814154920038789e-06,
+      "loss": 1.2721,
+      "step": 4247
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.17680587784903e-06,
+      "loss": 1.1803,
+      "step": 4248
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1722050015083752e-06,
+      "loss": 1.2556,
+      "step": 4249
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1676128651176578e-06,
+      "loss": 1.1844,
+      "step": 4250
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.163029470808572e-06,
+      "loss": 1.2093,
+      "step": 4251
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1584548207087498e-06,
+      "loss": 1.2231,
+      "step": 4252
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1538889169417654e-06,
+      "loss": 1.2151,
+      "step": 4253
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1493317616271327e-06,
+      "loss": 1.2287,
+      "step": 4254
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1447833568803036e-06,
+      "loss": 1.2537,
+      "step": 4255
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.140243704812667e-06,
+      "loss": 1.1861,
+      "step": 4256
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1357128075315572e-06,
+      "loss": 1.221,
+      "step": 4257
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1311906671402274e-06,
+      "loss": 1.2518,
+      "step": 4258
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.126677285737887e-06,
+      "loss": 1.2049,
+      "step": 4259
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.122172665419663e-06,
+      "loss": 1.2403,
+      "step": 4260
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1176768082766177e-06,
+      "loss": 1.1923,
+      "step": 4261
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1131897163957573e-06,
+      "loss": 1.1651,
+      "step": 4262
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1087113918600023e-06,
+      "loss": 1.1937,
+      "step": 4263
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1042418367482188e-06,
+      "loss": 1.3132,
+      "step": 4264
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0997810531351916e-06,
+      "loss": 1.2421,
+      "step": 4265
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0953290430916353e-06,
+      "loss": 1.2078,
+      "step": 4266
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0908858086841989e-06,
+      "loss": 1.2188,
+      "step": 4267
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0864513519754484e-06,
+      "loss": 1.2179,
+      "step": 4268
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.082025675023879e-06,
+      "loss": 1.2269,
+      "step": 4269
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.077608779883912e-06,
+      "loss": 1.2637,
+      "step": 4270
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0732006686058893e-06,
+      "loss": 1.241,
+      "step": 4271
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0688013432360811e-06,
+      "loss": 1.2497,
+      "step": 4272
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0644108058166692e-06,
+      "loss": 1.2513,
+      "step": 4273
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0600290583857631e-06,
+      "loss": 1.2318,
+      "step": 4274
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0556561029773914e-06,
+      "loss": 1.1844,
+      "step": 4275
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0512919416214995e-06,
+      "loss": 1.2313,
+      "step": 4276
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0469365763439532e-06,
+      "loss": 1.207,
+      "step": 4277
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0425900091665286e-06,
+      "loss": 1.291,
+      "step": 4278
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0382522421069274e-06,
+      "loss": 1.2355,
+      "step": 4279
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.033923277178759e-06,
+      "loss": 1.1848,
+      "step": 4280
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.029603116391551e-06,
+      "loss": 1.2407,
+      "step": 4281
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0252917617507374e-06,
+      "loss": 1.2639,
+      "step": 4282
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.020989215257675e-06,
+      "loss": 1.2355,
+      "step": 4283
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0166954789096194e-06,
+      "loss": 1.1912,
+      "step": 4284
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0124105546997521e-06,
+      "loss": 1.2359,
+      "step": 4285
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.008134444617146e-06,
+      "loss": 1.2382,
+      "step": 4286
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0038671506467934e-06,
+      "loss": 1.2216,
+      "step": 4287
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.996086747695966e-07,
+      "loss": 1.2366,
+      "step": 4288
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.953590189623563e-07,
+      "loss": 1.2396,
+      "step": 4289
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.911181851977792e-07,
+      "loss": 1.202,
+      "step": 4290
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.868861754444858e-07,
+      "loss": 1.2017,
+      "step": 4291
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.826629916669917e-07,
+      "loss": 1.2344,
+      "step": 4292
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.784486358257194e-07,
+      "loss": 1.2402,
+      "step": 4293
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.742431098769933e-07,
+      "loss": 1.2851,
+      "step": 4294
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.700464157730338e-07,
+      "loss": 1.1857,
+      "step": 4295
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.658585554619737e-07,
+      "loss": 1.2451,
+      "step": 4296
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.616795308878313e-07,
+      "loss": 1.2263,
+      "step": 4297
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.575093439905259e-07,
+      "loss": 1.236,
+      "step": 4298
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.533479967058867e-07,
+      "loss": 1.2494,
+      "step": 4299
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.491954909656242e-07,
+      "loss": 1.1863,
+      "step": 4300
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.450518286973542e-07,
+      "loss": 1.2832,
+      "step": 4301
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.409170118245803e-07,
+      "loss": 1.2413,
+      "step": 4302
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.36791042266707e-07,
+      "loss": 1.2495,
+      "step": 4303
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.326739219390246e-07,
+      "loss": 1.224,
+      "step": 4304
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.285656527527264e-07,
+      "loss": 1.23,
+      "step": 4305
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.244662366148826e-07,
+      "loss": 1.2561,
+      "step": 4306
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.203756754284665e-07,
+      "loss": 1.2677,
+      "step": 4307
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.162939710923324e-07,
+      "loss": 1.2469,
+      "step": 4308
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.122211255012292e-07,
+      "loss": 1.1916,
+      "step": 4309
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.081571405457912e-07,
+      "loss": 1.2088,
+      "step": 4310
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.041020181125315e-07,
+      "loss": 1.2659,
+      "step": 4311
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.000557600838666e-07,
+      "loss": 1.26,
+      "step": 4312
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.960183683380807e-07,
+      "loss": 1.2366,
+      "step": 4313
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.919898447493569e-07,
+      "loss": 1.2412,
+      "step": 4314
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.879701911877503e-07,
+      "loss": 1.2653,
+      "step": 4315
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.839594095191995e-07,
+      "loss": 1.1935,
+      "step": 4316
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.799575016055373e-07,
+      "loss": 1.2208,
+      "step": 4317
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.7596446930446e-07,
+      "loss": 1.2009,
+      "step": 4318
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.719803144695516e-07,
+      "loss": 1.2297,
+      "step": 4319
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.680050389502814e-07,
+      "loss": 1.258,
+      "step": 4320
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.640386445919847e-07,
+      "loss": 1.2318,
+      "step": 4321
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.600811332358861e-07,
+      "loss": 1.2014,
+      "step": 4322
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.561325067190762e-07,
+      "loss": 1.2076,
+      "step": 4323
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.521927668745244e-07,
+      "loss": 1.1566,
+      "step": 4324
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.482619155310812e-07,
+      "loss": 1.2294,
+      "step": 4325
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.443399545134623e-07,
+      "loss": 1.1682,
+      "step": 4326
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.404268856422626e-07,
+      "loss": 1.2093,
+      "step": 4327
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.365227107339447e-07,
+      "loss": 1.1701,
+      "step": 4328
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.326274316008475e-07,
+      "loss": 1.2414,
+      "step": 4329
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.287410500511739e-07,
+      "loss": 1.1983,
+      "step": 4330
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.248635678890049e-07,
+      "loss": 1.214,
+      "step": 4331
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.209949869142808e-07,
+      "loss": 1.2238,
+      "step": 4332
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.171353089228206e-07,
+      "loss": 1.2027,
+      "step": 4333
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.132845357062979e-07,
+      "loss": 1.2203,
+      "step": 4334
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.094426690522672e-07,
+      "loss": 1.2504,
+      "step": 4335
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.056097107441352e-07,
+      "loss": 1.2676,
+      "step": 4336
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.017856625611809e-07,
+      "loss": 1.2207,
+      "step": 4337
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.979705262785442e-07,
+      "loss": 1.2061,
+      "step": 4338
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.941643036672309e-07,
+      "loss": 1.2271,
+      "step": 4339
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.903669964941052e-07,
+      "loss": 1.2185,
+      "step": 4340
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.865786065218973e-07,
+      "loss": 1.2584,
+      "step": 4341
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.827991355091891e-07,
+      "loss": 1.2262,
+      "step": 4342
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.790285852104373e-07,
+      "loss": 1.2416,
+      "step": 4343
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.752669573759464e-07,
+      "loss": 1.2643,
+      "step": 4344
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.715142537518771e-07,
+      "loss": 1.1716,
+      "step": 4345
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.677704760802562e-07,
+      "loss": 1.2065,
+      "step": 4346
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.640356260989601e-07,
+      "loss": 1.2274,
+      "step": 4347
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.603097055417242e-07,
+      "loss": 1.24,
+      "step": 4348
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.565927161381403e-07,
+      "loss": 1.2557,
+      "step": 4349
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.528846596136485e-07,
+      "loss": 1.2253,
+      "step": 4350
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.491855376895519e-07,
+      "loss": 1.2702,
+      "step": 4351
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.454953520829899e-07,
+      "loss": 1.2337,
+      "step": 4352
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.418141045069727e-07,
+      "loss": 1.2667,
+      "step": 4353
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.381417966703508e-07,
+      "loss": 1.243,
+      "step": 4354
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.344784302778274e-07,
+      "loss": 1.2792,
+      "step": 4355
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.308240070299489e-07,
+      "loss": 1.206,
+      "step": 4356
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.271785286231204e-07,
+      "loss": 1.2438,
+      "step": 4357
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.235419967495883e-07,
+      "loss": 1.2203,
+      "step": 4358
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.199144130974489e-07,
+      "loss": 1.2486,
+      "step": 4359
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.16295779350642e-07,
+      "loss": 1.2434,
+      "step": 4360
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.126860971889527e-07,
+      "loss": 1.2417,
+      "step": 4361
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.090853682880161e-07,
+      "loss": 1.2403,
+      "step": 4362
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.054935943193042e-07,
+      "loss": 1.2814,
+      "step": 4363
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.019107769501366e-07,
+      "loss": 1.2591,
+      "step": 4364
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.983369178436739e-07,
+      "loss": 1.3018,
+      "step": 4365
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.947720186589158e-07,
+      "loss": 1.2351,
+      "step": 4366
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.912160810507096e-07,
+      "loss": 1.2228,
+      "step": 4367
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.876691066697349e-07,
+      "loss": 1.2691,
+      "step": 4368
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.841310971625103e-07,
+      "loss": 1.2701,
+      "step": 4369
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.806020541714042e-07,
+      "loss": 1.2439,
+      "step": 4370
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.770819793346084e-07,
+      "loss": 1.2196,
+      "step": 4371
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.735708742861624e-07,
+      "loss": 1.2263,
+      "step": 4372
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.700687406559359e-07,
+      "loss": 1.2102,
+      "step": 4373
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.665755800696305e-07,
+      "loss": 1.2618,
+      "step": 4374
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.63091394148796e-07,
+      "loss": 1.1802,
+      "step": 4375
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.596161845108006e-07,
+      "loss": 1.2205,
+      "step": 4376
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.561499527688586e-07,
+      "loss": 1.1975,
+      "step": 4377
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.526927005320072e-07,
+      "loss": 1.1822,
+      "step": 4378
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.492444294051204e-07,
+      "loss": 1.2436,
+      "step": 4379
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.458051409889021e-07,
+      "loss": 1.2051,
+      "step": 4380
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.423748368798843e-07,
+      "loss": 1.2292,
+      "step": 4381
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.38953518670431e-07,
+      "loss": 1.2116,
+      "step": 4382
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.355411879487339e-07,
+      "loss": 1.2199,
+      "step": 4383
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.321378462988148e-07,
+      "loss": 1.2384,
+      "step": 4384
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.287434953005145e-07,
+      "loss": 1.2618,
+      "step": 4385
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.253581365295148e-07,
+      "loss": 1.2689,
+      "step": 4386
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.219817715573073e-07,
+      "loss": 1.2113,
+      "step": 4387
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.18614401951223e-07,
+      "loss": 1.2616,
+      "step": 4388
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.152560292744091e-07,
+      "loss": 1.2255,
+      "step": 4389
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.119066550858321e-07,
+      "loss": 1.2419,
+      "step": 4390
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.085662809402926e-07,
+      "loss": 1.2623,
+      "step": 4391
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.052349083884057e-07,
+      "loss": 1.2152,
+      "step": 4392
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.019125389766123e-07,
+      "loss": 1.2353,
+      "step": 4393
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.985991742471698e-07,
+      "loss": 1.2339,
+      "step": 4394
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.952948157381566e-07,
+      "loss": 1.1722,
+      "step": 4395
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.919994649834748e-07,
+      "loss": 1.2445,
+      "step": 4396
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.887131235128385e-07,
+      "loss": 1.2186,
+      "step": 4397
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.854357928517806e-07,
+      "loss": 1.2297,
+      "step": 4398
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.821674745216599e-07,
+      "loss": 1.2235,
+      "step": 4399
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.789081700396381e-07,
+      "loss": 1.1534,
+      "step": 4400
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.756578809187008e-07,
+      "loss": 1.2411,
+      "step": 4401
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.724166086676542e-07,
+      "loss": 1.2196,
+      "step": 4402
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.691843547911013e-07,
+      "loss": 1.2477,
+      "step": 4403
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.6596112078948e-07,
+      "loss": 1.216,
+      "step": 4404
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.627469081590242e-07,
+      "loss": 1.2527,
+      "step": 4405
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.595417183917851e-07,
+      "loss": 1.2296,
+      "step": 4406
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.563455529756301e-07,
+      "loss": 1.2075,
+      "step": 4407
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.531584133942325e-07,
+      "loss": 1.2425,
+      "step": 4408
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.499803011270776e-07,
+      "loss": 1.2671,
+      "step": 4409
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.468112176494633e-07,
+      "loss": 1.2547,
+      "step": 4410
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.43651164432486e-07,
+      "loss": 1.2541,
+      "step": 4411
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.405001429430634e-07,
+      "loss": 1.195,
+      "step": 4412
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.373581546439077e-07,
+      "loss": 1.2237,
+      "step": 4413
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.342252009935522e-07,
+      "loss": 1.2363,
+      "step": 4414
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.311012834463247e-07,
+      "loss": 1.2104,
+      "step": 4415
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.279864034523586e-07,
+      "loss": 1.2381,
+      "step": 4416
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.248805624576037e-07,
+      "loss": 1.2374,
+      "step": 4417
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.217837619038002e-07,
+      "loss": 1.2506,
+      "step": 4418
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.186960032284983e-07,
+      "loss": 1.2166,
+      "step": 4419
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.156172878650489e-07,
+      "loss": 1.2131,
+      "step": 4420
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.125476172426092e-07,
+      "loss": 1.2366,
+      "step": 4421
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.094869927861323e-07,
+      "loss": 1.2112,
+      "step": 4422
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.064354159163754e-07,
+      "loss": 1.2071,
+      "step": 4423
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.033928880498917e-07,
+      "loss": 1.3022,
+      "step": 4424
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.003594105990384e-07,
+      "loss": 1.21,
+      "step": 4425
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.973349849719733e-07,
+      "loss": 1.1817,
+      "step": 4426
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.943196125726446e-07,
+      "loss": 1.2437,
+      "step": 4427
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.913132948008037e-07,
+      "loss": 1.1759,
+      "step": 4428
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.883160330519965e-07,
+      "loss": 1.2422,
+      "step": 4429
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.853278287175677e-07,
+      "loss": 1.2204,
+      "step": 4430
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.823486831846547e-07,
+      "loss": 1.2376,
+      "step": 4431
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.793785978361887e-07,
+      "loss": 1.2706,
+      "step": 4432
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.764175740509025e-07,
+      "loss": 1.2328,
+      "step": 4433
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.7346561320330997e-07,
+      "loss": 1.2333,
+      "step": 4434
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.7052271666373053e-07,
+      "loss": 1.1884,
+      "step": 4435
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.675888857982669e-07,
+      "loss": 1.2321,
+      "step": 4436
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.646641219688186e-07,
+      "loss": 1.2524,
+      "step": 4437
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.617484265330752e-07,
+      "loss": 1.2138,
+      "step": 4438
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.588418008445161e-07,
+      "loss": 1.2546,
+      "step": 4439
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.5594424625240887e-07,
+      "loss": 1.2357,
+      "step": 4440
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.5305576410181293e-07,
+      "loss": 1.2278,
+      "step": 4441
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.5017635573357366e-07,
+      "loss": 1.2714,
+      "step": 4442
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.4730602248432843e-07,
+      "loss": 1.2129,
+      "step": 4443
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.444447656864981e-07,
+      "loss": 1.2472,
+      "step": 4444
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.4159258666828907e-07,
+      "loss": 1.2369,
+      "step": 4445
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.3874948675370233e-07,
+      "loss": 1.2305,
+      "step": 4446
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.3591546726250877e-07,
+      "loss": 1.234,
+      "step": 4447
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.3309052951028275e-07,
+      "loss": 1.2171,
+      "step": 4448
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.302746748083664e-07,
+      "loss": 1.1726,
+      "step": 4449
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.2746790446389853e-07,
+      "loss": 1.2404,
+      "step": 4450
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.2467021977978806e-07,
+      "loss": 1.1928,
+      "step": 4451
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.218816220547406e-07,
+      "loss": 1.2398,
+      "step": 4452
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.1910211258322954e-07,
+      "loss": 1.2147,
+      "step": 4453
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.1633169265552274e-07,
+      "loss": 1.2187,
+      "step": 4454
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.13570363557656e-07,
+      "loss": 1.2409,
+      "step": 4455
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.108181265714528e-07,
+      "loss": 1.2315,
+      "step": 4456
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.0807498297451786e-07,
+      "loss": 1.2053,
+      "step": 4457
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.053409340402259e-07,
+      "loss": 1.1859,
+      "step": 4458
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.026159810377417e-07,
+      "loss": 1.1849,
+      "step": 4459
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.999001252319934e-07,
+      "loss": 1.2131,
+      "step": 4460
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.971933678836992e-07,
+      "loss": 1.283,
+      "step": 4461
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.944957102493474e-07,
+      "loss": 1.2054,
+      "step": 4462
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.918071535812007e-07,
+      "loss": 1.2272,
+      "step": 4463
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.8912769912730297e-07,
+      "loss": 1.2537,
+      "step": 4464
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.864573481314682e-07,
+      "loss": 1.2086,
+      "step": 4465
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.837961018332825e-07,
+      "loss": 1.2417,
+      "step": 4466
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.811439614681156e-07,
+      "loss": 1.2355,
+      "step": 4467
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.7850092826709817e-07,
+      "loss": 1.2388,
+      "step": 4468
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.758670034571399e-07,
+      "loss": 1.222,
+      "step": 4469
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.7324218826092053e-07,
+      "loss": 1.1949,
+      "step": 4470
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.7062648389689204e-07,
+      "loss": 1.2616,
+      "step": 4471
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.680198915792765e-07,
+      "loss": 1.2673,
+      "step": 4472
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.654224125180661e-07,
+      "loss": 1.1847,
+      "step": 4473
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.628340479190229e-07,
+      "loss": 1.253,
+      "step": 4474
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.602547989836769e-07,
+      "loss": 1.2759,
+      "step": 4475
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.5768466690933036e-07,
+      "loss": 1.2087,
+      "step": 4476
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.551236528890445e-07,
+      "loss": 1.2671,
+      "step": 4477
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.5257175811166166e-07,
+      "loss": 1.2362,
+      "step": 4478
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.500289837617765e-07,
+      "loss": 1.2305,
+      "step": 4479
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.474953310197604e-07,
+      "loss": 1.2287,
+      "step": 4480
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4497080106174806e-07,
+      "loss": 1.2224,
+      "step": 4481
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.424553950596332e-07,
+      "loss": 1.2744,
+      "step": 4482
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.3994911418108176e-07,
+      "loss": 1.2061,
+      "step": 4483
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.374519595895209e-07,
+      "loss": 1.2135,
+      "step": 4484
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.3496393244414114e-07,
+      "loss": 1.2378,
+      "step": 4485
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.324850338998964e-07,
+      "loss": 1.244,
+      "step": 4486
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.300152651075039e-07,
+      "loss": 1.2142,
+      "step": 4487
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.2755462721344e-07,
+      "loss": 1.2057,
+      "step": 4488
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.251031213599465e-07,
+      "loss": 1.2561,
+      "step": 4489
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.2266074868501976e-07,
+      "loss": 1.2426,
+      "step": 4490
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.2022751032242396e-07,
+      "loss": 1.2202,
+      "step": 4491
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.178034074016778e-07,
+      "loss": 1.2431,
+      "step": 4492
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.1538844104806343e-07,
+      "loss": 1.2423,
+      "step": 4493
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.1298261238261964e-07,
+      "loss": 1.2351,
+      "step": 4494
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.105859225221397e-07,
+      "loss": 1.2049,
+      "step": 4495
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.0819837257918037e-07,
+      "loss": 1.2546,
+      "step": 4496
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.058199636620529e-07,
+      "loss": 1.1829,
+      "step": 4497
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.034506968748274e-07,
+      "loss": 1.2594,
+      "step": 4498
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.010905733173264e-07,
+      "loss": 1.2462,
+      "step": 4499
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.987395940851312e-07,
+      "loss": 1.239,
+      "step": 4500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.9639776026957777e-07,
+      "loss": 1.2714,
+      "step": 4501
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.9406507295775657e-07,
+      "loss": 1.2536,
+      "step": 4502
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.91741533232508e-07,
+      "loss": 1.2272,
+      "step": 4503
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.894271421724359e-07,
+      "loss": 1.2041,
+      "step": 4504
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.871219008518877e-07,
+      "loss": 1.2234,
+      "step": 4505
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8482581034096733e-07,
+      "loss": 1.2461,
+      "step": 4506
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.825388717055311e-07,
+      "loss": 1.2007,
+      "step": 4507
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8026108600718746e-07,
+      "loss": 1.2223,
+      "step": 4508
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.7799245430329526e-07,
+      "loss": 1.2345,
+      "step": 4509
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.7573297764696085e-07,
+      "loss": 1.2377,
+      "step": 4510
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.7348265708704745e-07,
+      "loss": 1.1932,
+      "step": 4511
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.7124149366816177e-07,
+      "loss": 1.2411,
+      "step": 4512
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.690094884306649e-07,
+      "loss": 1.2273,
+      "step": 4513
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.667866424106591e-07,
+      "loss": 1.2573,
+      "step": 4514
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6457295664000573e-07,
+      "loss": 1.2285,
+      "step": 4515
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.623684321463049e-07,
+      "loss": 1.2345,
+      "step": 4516
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6017306995290926e-07,
+      "loss": 1.2686,
+      "step": 4517
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.579868710789124e-07,
+      "loss": 1.2229,
+      "step": 4518
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5580983653916035e-07,
+      "loss": 1.2112,
+      "step": 4519
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5364196734424475e-07,
+      "loss": 1.2086,
+      "step": 4520
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.514832645004939e-07,
+      "loss": 1.2545,
+      "step": 4521
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.493337290099973e-07,
+      "loss": 1.2147,
+      "step": 4522
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.471933618705702e-07,
+      "loss": 1.2331,
+      "step": 4523
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4506216407578665e-07,
+      "loss": 1.2037,
+      "step": 4524
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.429401366149553e-07,
+      "loss": 1.2079,
+      "step": 4525
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4082728047313487e-07,
+      "loss": 1.2442,
+      "step": 4526
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.3872359663111856e-07,
+      "loss": 1.2323,
+      "step": 4527
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.3662908606544964e-07,
+      "loss": 1.2278,
+      "step": 4528
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.34543749748406e-07,
+      "loss": 1.279,
+      "step": 4529
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.3246758864801544e-07,
+      "loss": 1.2132,
+      "step": 4530
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.304006037280404e-07,
+      "loss": 1.2389,
+      "step": 4531
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.2834279594798002e-07,
+      "loss": 1.2062,
+      "step": 4532
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.2629416626308353e-07,
+      "loss": 1.1694,
+      "step": 4533
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.2425471562433466e-07,
+      "loss": 1.2011,
+      "step": 4534
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.222244449784494e-07,
+      "loss": 1.1841,
+      "step": 4535
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.2020335526789616e-07,
+      "loss": 1.2362,
+      "step": 4536
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1819144743086883e-07,
+      "loss": 1.2204,
+      "step": 4537
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1618872240130928e-07,
+      "loss": 1.2544,
+      "step": 4538
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1419518110888938e-07,
+      "loss": 1.2338,
+      "step": 4539
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1221082447901774e-07,
+      "loss": 1.2449,
+      "step": 4540
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1023565343284425e-07,
+      "loss": 1.2354,
+      "step": 4541
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.082696688872554e-07,
+      "loss": 1.2078,
+      "step": 4542
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.063128717548657e-07,
+      "loss": 1.2149,
+      "step": 4543
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.043652629440307e-07,
+      "loss": 1.2556,
+      "step": 4544
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0242684335884056e-07,
+      "loss": 1.2607,
+      "step": 4545
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0049761389911772e-07,
+      "loss": 1.2898,
+      "step": 4546
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9857757546041912e-07,
+      "loss": 1.2345,
+      "step": 4547
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9666672893403627e-07,
+      "loss": 1.2257,
+      "step": 4548
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9476507520699518e-07,
+      "loss": 1.2238,
+      "step": 4549
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.928726151620497e-07,
+      "loss": 1.2325,
+      "step": 4550
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9098934967768823e-07,
+      "loss": 1.2727,
+      "step": 4551
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.891152796281337e-07,
+      "loss": 1.2301,
+      "step": 4552
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8725040588333466e-07,
+      "loss": 1.1936,
+      "step": 4553
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.853947293089764e-07,
+      "loss": 1.2217,
+      "step": 4554
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8354825076647432e-07,
+      "loss": 1.2005,
+      "step": 4555
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.817109711129672e-07,
+      "loss": 1.1929,
+      "step": 4556
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7988289120133507e-07,
+      "loss": 1.2421,
+      "step": 4557
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7806401188017463e-07,
+      "loss": 1.2203,
+      "step": 4558
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7625433399382386e-07,
+      "loss": 1.2239,
+      "step": 4559
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7445385838234185e-07,
+      "loss": 1.223,
+      "step": 4560
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7266258588151562e-07,
+      "loss": 1.1813,
+      "step": 4561
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7088051732286448e-07,
+      "loss": 1.2105,
+      "step": 4562
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6910765353363334e-07,
+      "loss": 1.2597,
+      "step": 4563
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6734399533679057e-07,
+      "loss": 1.237,
+      "step": 4564
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6558954355103686e-07,
+      "loss": 1.2297,
+      "step": 4565
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6384429899079624e-07,
+      "loss": 1.2204,
+      "step": 4566
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6210826246622068e-07,
+      "loss": 1.1876,
+      "step": 4567
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.603814347831856e-07,
+      "loss": 1.2152,
+      "step": 4568
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.586638167432919e-07,
+      "loss": 1.2398,
+      "step": 4569
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5695540914386632e-07,
+      "loss": 1.2265,
+      "step": 4570
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.552562127779611e-07,
+      "loss": 1.1975,
+      "step": 4571
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5356622843434533e-07,
+      "loss": 1.1959,
+      "step": 4572
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.51885456897527e-07,
+      "loss": 1.2192,
+      "step": 4573
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5021389894771753e-07,
+      "loss": 1.2153,
+      "step": 4574
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4855155536087184e-07,
+      "loss": 1.2368,
+      "step": 4575
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4689842690865042e-07,
+      "loss": 1.2154,
+      "step": 4576
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4525451435844608e-07,
+      "loss": 1.2416,
+      "step": 4577
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.436198184733706e-07,
+      "loss": 1.1973,
+      "step": 4578
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4199434001225697e-07,
+      "loss": 1.2245,
+      "step": 4579
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4037807972966167e-07,
+      "loss": 1.2799,
+      "step": 4580
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.387710383758556e-07,
+      "loss": 1.2327,
+      "step": 4581
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3717321669683981e-07,
+      "loss": 1.2359,
+      "step": 4582
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3558461543432767e-07,
+      "loss": 1.2137,
+      "step": 4583
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3400523532575592e-07,
+      "loss": 1.2329,
+      "step": 4584
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.324350771042804e-07,
+      "loss": 1.2091,
+      "step": 4585
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3087414149877574e-07,
+      "loss": 1.2304,
+      "step": 4586
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.2932242923383575e-07,
+      "loss": 1.2547,
+      "step": 4587
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.277799410297731e-07,
+      "loss": 1.2672,
+      "step": 4588
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.26246677602615e-07,
+      "loss": 1.2167,
+      "step": 4589
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2472263966411214e-07,
+      "loss": 1.2341,
+      "step": 4590
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2320782792173192e-07,
+      "loss": 1.2091,
+      "step": 4591
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2170224307865185e-07,
+      "loss": 1.2378,
+      "step": 4592
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2020588583377513e-07,
+      "loss": 1.2021,
+      "step": 4593
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.187187568817172e-07,
+      "loss": 1.2016,
+      "step": 4594
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1724085691280806e-07,
+      "loss": 1.2125,
+      "step": 4595
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1577218661309896e-07,
+      "loss": 1.2236,
+      "step": 4596
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1431274666435121e-07,
+      "loss": 1.161,
+      "step": 4597
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1286253774404288e-07,
+      "loss": 1.2346,
+      "step": 4598
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.11421560525371e-07,
+      "loss": 1.2394,
+      "step": 4599
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0998981567724276e-07,
+      "loss": 1.2812,
+      "step": 4600
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0856730386427983e-07,
+      "loss": 1.2264,
+      "step": 4601
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0715402574681843e-07,
+      "loss": 1.2406,
+      "step": 4602
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0574998198090935e-07,
+      "loss": 1.2552,
+      "step": 4603
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0435517321831568e-07,
+      "loss": 1.1945,
+      "step": 4604
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0296960010651725e-07,
+      "loss": 1.1916,
+      "step": 4605
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0159326328869734e-07,
+      "loss": 1.1895,
+      "step": 4606
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0022616340376489e-07,
+      "loss": 1.2183,
+      "step": 4607
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.886830108632784e-08,
+      "loss": 1.209,
+      "step": 4608
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.751967696671749e-08,
+      "loss": 1.2673,
+      "step": 4609
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.618029167096865e-08,
+      "loss": 1.271,
+      "step": 4610
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.485014582083063e-08,
+      "loss": 1.1878,
+      "step": 4611
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.352924003376285e-08,
+      "loss": 1.2163,
+      "step": 4612
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.221757492293704e-08,
+      "loss": 1.2079,
+      "step": 4613
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.091515109723281e-08,
+      "loss": 1.2322,
+      "step": 4614
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.96219691612421e-08,
+      "loss": 1.213,
+      "step": 4615
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.833802971526472e-08,
+      "loss": 1.2398,
+      "step": 4616
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.706333335531503e-08,
+      "loss": 1.2366,
+      "step": 4617
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.579788067310858e-08,
+      "loss": 1.2867,
+      "step": 4618
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.454167225607768e-08,
+      "loss": 1.2004,
+      "step": 4619
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.32947086873559e-08,
+      "loss": 1.235,
+      "step": 4620
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.205699054579575e-08,
+      "loss": 1.2393,
+      "step": 4621
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.082851840594652e-08,
+      "loss": 1.2185,
+      "step": 4622
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.960929283807429e-08,
+      "loss": 1.2763,
+      "step": 4623
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.83993144081463e-08,
+      "loss": 1.1988,
+      "step": 4624
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.719858367784216e-08,
+      "loss": 1.2469,
+      "step": 4625
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.600710120454491e-08,
+      "loss": 1.2057,
+      "step": 4626
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.482486754134765e-08,
+      "loss": 1.2969,
+      "step": 4627
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.365188323704919e-08,
+      "loss": 1.2414,
+      "step": 4628
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.248814883615174e-08,
+      "loss": 1.2522,
+      "step": 4629
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.133366487886762e-08,
+      "loss": 1.2131,
+      "step": 4630
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.018843190111479e-08,
+      "loss": 1.2519,
+      "step": 4631
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.90524504345147e-08,
+      "loss": 1.2755,
+      "step": 4632
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.792572100639661e-08,
+      "loss": 1.2914,
+      "step": 4633
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.680824413979103e-08,
+      "loss": 1.2579,
+      "step": 4634
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.570002035343636e-08,
+      "loss": 1.2256,
+      "step": 4635
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.460105016177887e-08,
+      "loss": 1.2272,
+      "step": 4636
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.351133407495936e-08,
+      "loss": 1.2768,
+      "step": 4637
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.24308725988354e-08,
+      "loss": 1.1901,
+      "step": 4638
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.135966623495915e-08,
+      "loss": 1.2421,
+      "step": 4639
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.029771548058838e-08,
+      "loss": 1.2004,
+      "step": 4640
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.924502082868655e-08,
+      "loss": 1.2288,
+      "step": 4641
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.820158276792054e-08,
+      "loss": 1.2435,
+      "step": 4642
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.716740178266067e-08,
+      "loss": 1.2091,
+      "step": 4643
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.614247835297404e-08,
+      "loss": 1.1938,
+      "step": 4644
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.512681295463784e-08,
+      "loss": 1.2725,
+      "step": 4645
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.4120406059128274e-08,
+      "loss": 1.2059,
+      "step": 4646
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.312325813362274e-08,
+      "loss": 1.2319,
+      "step": 4647
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.21353696410043e-08,
+      "loss": 1.1882,
+      "step": 4648
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.11567410398528e-08,
+      "loss": 1.221,
+      "step": 4649
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.0187372784453734e-08,
+      "loss": 1.2138,
+      "step": 4650
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.922726532479383e-08,
+      "loss": 1.2108,
+      "step": 4651
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.827641910655656e-08,
+      "loss": 1.2147,
+      "step": 4652
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.7334834571128866e-08,
+      "loss": 1.2249,
+      "step": 4653
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.640251215560332e-08,
+      "loss": 1.1814,
+      "step": 4654
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.547945229276263e-08,
+      "loss": 1.2164,
+      "step": 4655
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.456565541109958e-08,
+      "loss": 1.187,
+      "step": 4656
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.366112193480154e-08,
+      "loss": 1.1888,
+      "step": 4657
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.276585228375485e-08,
+      "loss": 1.2154,
+      "step": 4658
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.187984687355151e-08,
+      "loss": 1.1806,
+      "step": 4659
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.100310611547809e-08,
+      "loss": 1.2458,
+      "step": 4660
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.013563041652013e-08,
+      "loss": 1.2249,
+      "step": 4661
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.927742017936664e-08,
+      "loss": 1.2553,
+      "step": 4662
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.8428475802398944e-08,
+      "loss": 1.2188,
+      "step": 4663
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.7588797679706245e-08,
+      "loss": 1.2438,
+      "step": 4664
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.6758386201065645e-08,
+      "loss": 1.2461,
+      "step": 4665
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.5937241751962115e-08,
+      "loss": 1.2565,
+      "step": 4666
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.5125364713572976e-08,
+      "loss": 1.2366,
+      "step": 4667
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.4322755462774525e-08,
+      "loss": 1.1382,
+      "step": 4668
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3529414372142074e-08,
+      "loss": 1.2868,
+      "step": 4669
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.2745341809949923e-08,
+      "loss": 1.233,
+      "step": 4670
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1970538140166927e-08,
+      "loss": 1.2067,
+      "step": 4671
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1205003722460935e-08,
+      "loss": 1.2252,
+      "step": 4672
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0448738912196574e-08,
+      "loss": 1.2462,
+      "step": 4673
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9701744060435246e-08,
+      "loss": 1.2097,
+      "step": 4674
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8964019513935126e-08,
+      "loss": 1.2165,
+      "step": 4675
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8235565615151172e-08,
+      "loss": 1.1683,
+      "step": 4676
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7516382702235112e-08,
+      "loss": 1.1979,
+      "step": 4677
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6806471109037668e-08,
+      "loss": 1.2402,
+      "step": 4678
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6105831165099683e-08,
+      "loss": 1.2255,
+      "step": 4679
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.541446319566321e-08,
+      "loss": 1.2091,
+      "step": 4680
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.473236752166264e-08,
+      "loss": 1.209,
+      "step": 4681
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4059544459731356e-08,
+      "loss": 1.1749,
+      "step": 4682
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3395994322199522e-08,
+      "loss": 1.18,
+      "step": 4683
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2741717417085196e-08,
+      "loss": 1.2378,
+      "step": 4684
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2096714048109867e-08,
+      "loss": 1.2202,
+      "step": 4685
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1460984514685145e-08,
+      "loss": 1.2915,
+      "step": 4686
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.083452911192163e-08,
+      "loss": 1.2164,
+      "step": 4687
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.021734813062226e-08,
+      "loss": 1.2087,
+      "step": 4688
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.960944185728675e-08,
+      "loss": 1.1995,
+      "step": 4689
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9010810574102702e-08,
+      "loss": 1.1979,
+      "step": 4690
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.842145455896338e-08,
+      "loss": 1.2474,
+      "step": 4691
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7841374085447728e-08,
+      "loss": 1.2469,
+      "step": 4692
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.727056942283367e-08,
+      "loss": 1.2645,
+      "step": 4693
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6709040836089262e-08,
+      "loss": 1.2161,
+      "step": 4694
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6156788585879325e-08,
+      "loss": 1.2441,
+      "step": 4695
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5613812928563233e-08,
+      "loss": 1.2384,
+      "step": 4696
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.508011411619048e-08,
+      "loss": 1.2884,
+      "step": 4697
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4555692396509557e-08,
+      "loss": 1.2231,
+      "step": 4698
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4040548012956844e-08,
+      "loss": 1.1984,
+      "step": 4699
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3534681204665502e-08,
+      "loss": 1.2186,
+      "step": 4700
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3038092206461017e-08,
+      "loss": 1.2536,
+      "step": 4701
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2550781248863442e-08,
+      "loss": 1.2366,
+      "step": 4702
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2072748558082936e-08,
+      "loss": 1.2491,
+      "step": 4703
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1603994356026437e-08,
+      "loss": 1.2115,
+      "step": 4704
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1144518860290998e-08,
+      "loss": 1.2219,
+      "step": 4705
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0694322284166003e-08,
+      "loss": 1.2137,
+      "step": 4706
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0253404836637615e-08,
+      "loss": 1.1968,
+      "step": 4707
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.821766722379888e-09,
+      "loss": 1.2349,
+      "step": 4708
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.399408141761434e-09,
+      "loss": 1.185,
+      "step": 4709
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.9863292908432e-09,
+      "loss": 1.258,
+      "step": 4710
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.58253036137846e-09,
+      "loss": 1.2634,
+      "step": 4711
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.188011540812834e-09,
+      "loss": 1.2651,
+      "step": 4712
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.80277301228205e-09,
+      "loss": 1.2436,
+      "step": 4713
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.426814954618611e-09,
+      "loss": 1.2377,
+      "step": 4714
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.060137542340695e-09,
+      "loss": 1.2328,
+      "step": 4715
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.702740945663256e-09,
+      "loss": 1.1509,
+      "step": 4716
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.35462533049358e-09,
+      "loss": 1.2448,
+      "step": 4717
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.0157908584246305e-09,
+      "loss": 1.2704,
+      "step": 4718
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.686237686746143e-09,
+      "loss": 1.2092,
+      "step": 4719
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.36596596844019e-09,
+      "loss": 1.2177,
+      "step": 4720
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.054975852176736e-09,
+      "loss": 1.2601,
+      "step": 4721
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.7532674823203e-09,
+      "loss": 1.2601,
+      "step": 4722
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.4608409989232995e-09,
+      "loss": 1.2555,
+      "step": 4723
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.17769653773048e-09,
+      "loss": 1.2481,
+      "step": 4724
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.903834230183368e-09,
+      "loss": 1.2643,
+      "step": 4725
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.639254203406939e-09,
+      "loss": 1.2755,
+      "step": 4726
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.383956580218506e-09,
+      "loss": 1.1891,
+      "step": 4727
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.1379414791343766e-09,
+      "loss": 1.2494,
+      "step": 4728
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9012090143498704e-09,
+      "loss": 1.2551,
+      "step": 4729
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6737592957615243e-09,
+      "loss": 1.1741,
+      "step": 4730
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4555924289493272e-09,
+      "loss": 1.2429,
+      "step": 4731
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2467085151900436e-09,
+      "loss": 1.2265,
+      "step": 4732
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.047107651446112e-09,
+      "loss": 1.2828,
+      "step": 4733
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8567899303767457e-09,
+      "loss": 1.2395,
+      "step": 4734
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6757554403223907e-09,
+      "loss": 1.2633,
+      "step": 4735
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5040042653269304e-09,
+      "loss": 1.2112,
+      "step": 4736
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3415364851132595e-09,
+      "loss": 1.2111,
+      "step": 4737
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.188352175103269e-09,
+      "loss": 1.2249,
+      "step": 4738
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0444514064023025e-09,
+      "loss": 1.2669,
+      "step": 4739
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.098342458102593e-10,
+      "loss": 1.2549,
+      "step": 4740
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.845007558193729e-10,
+      "loss": 1.243,
+      "step": 4741
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.684509946075501e-10,
+      "loss": 1.1938,
+      "step": 4742
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.616850160494736e-10,
+      "loss": 1.2334,
+      "step": 4743
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.642028697010581e-10,
+      "loss": 1.2111,
+      "step": 4744
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.760046008172147e-10,
+      "loss": 1.2216,
+      "step": 4745
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.97090250340748e-10,
+      "loss": 1.2322,
+      "step": 4746
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2745985490235656e-10,
+      "loss": 1.1637,
+      "step": 4747
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6711344682507346e-10,
+      "loss": 1.2659,
+      "step": 4748
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.160510541220461e-10,
+      "loss": 1.2167,
+      "step": 4749
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.427270049653601e-11,
+      "loss": 1.2163,
+      "step": 4750
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.1778405341919016e-11,
+      "loss": 1.2346,
+      "step": 4751
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8568183743905566e-11,
+      "loss": 1.2219,
+      "step": 4752
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6420464738794465e-12,
+      "loss": 1.2163,
+      "step": 4753
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 1.1564,
+      "step": 4754
+    },
+    {
+      "epoch": 1.0,
+      "step": 4754,
+      "total_flos": 1.1295276544628886e+23,
+      "train_loss": 1.2829506933864647,
+      "train_runtime": 87797.1342,
+      "train_samples_per_second": 6.931,
+      "train_steps_per_second": 0.054
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 4754,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 5000,
+  "total_flos": 1.1295276544628886e+23,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}